The Impact of A Family Empowerment Intervention On Juvenile Offender Heavy Drinking

MARCEL DEKKER, INC.
270 MADISON AVENUE NEW YORK, NY 10016

2002 Marcel Dekker, Inc. All rights reserved. This material may not be used or reproduced in any form without the express written permission of Marcel Dekker, Inc.
SUBSTANCE USE & MISUSE

Vol. 37, No. 11, pp. 13591390, 2002
THE IMPACT OF A FAMILY

EMPOWERMENT INTERVENTION
ON JUVENILE OFFENDER HEAVY
DRINKING: A LATENT GROWTH
MODEL ANALYSIS*
Richard Dembo, Ph.D.,1,y Werner Wothke, Ph.D.,2

Stephen Livingston, B.A.,1 and
James Schmeidler, Ph.D.3
1
Department of Criminology, University of South
Florida, 4202 E. Fowler Avenue, Tampa, FL 33620
2
SmallWaters Corporation, 1507 E. 53rd St.,
Suite 452, Chicago, IL 60615
3
Departments of Psychiatry and Biomathematical
Sciences, Mt. Sinai School of Medicine,
One Gustave L. Levy Place, New York, NY 10029
ABSTRACT
We report the results of a growth model analysis of the impact

of a Family Empowerment Intervention (FEI) on the heavy
drinking over a 36-month follow-up period among youths
*Preparation of this manuscript was supported by Grant # 1-R01-DA08707, funded

by the National Institute on Drug Abuse. We are grateful for their support.
However, the research results reported and the views expressed in the paper do
not necessarily imply any policy or research endorsement by our funding agency.
y
Corresponding author. Fax: 813-933-1146.
1359
DOI: 10.1081/JA-120014082 1082-6084 (Print); 1532-2491 (Online)

Copyright & 2002 by Marcel Dekker, Inc. www.dekker.com
MARCEL DEKKER, INC. 270 MADISON AVENUE NEW YORK, NY 10016
1360 DEMBO ET AL.
processed at the Hillsborough County Juvenile Assessment

Center. Families involved in the project were randomly
assigned to either receive an Extended Services Intervention
(ESI) or the FEI. Families in the ESI group received monthly
phone contacts and, if indicated, referral information; FEI
families received three one-hour, home-based meetings per
week for approximately 10 weeks from a clinician-trained
paraprofessional. By seeking to improve family functioning
by empowering parents, it was hypothesized that target
youths behavior and psychosocial functioning would improve.
Although the dierence between FEI and ESI was not signi-
cant, the reported frequency of getting very high or drunk on
alcohol declined more over time for FEI completers than FEI
noncompleters. The results provide support for the impact of
the FEI services.
Key Words: Family empowerment intervention; Alcohol

use; High risk youth
INTRODUCTION
By any criterion, alcohol use among U.S. youths remains at an

unacceptably high level. For example, the Substance Abuse and Mental
Health Services Administration (SAMSHA) funded 1999 household
survey on drug use found that current (past 30 days) alcohol use among
youths increased from 3.9% among 1213 year-olds to 50% among persons
18 to 20 years of age.[1] Further, although it is illegal for persons under 21 to
consume alcohol, an estimated 10.4 million persons 12 to 20 years of age
were current drinkers: 6.8 million were binge drinkers (consuming 5 or more
drinks on some occasion during the 30-day period prior to their interviews),
and 2.1 million were heavy drinkers (having 5 or more drinks on one
occasion 5 or more days during the past 30 days).[1]
Related results from the 2000 Monitoring the Future Study indicate
that alcohol use among U.S. youths has remained stable at fairly high
levels during the past several years. Current use was reported by 22%,
41%, and 50% of youths in grades eight, 10, and 12, respectively. In addition,
a slight uptrend in binge drinking was found for youths in lower grades, but
these rates were in line with those reported two or three years ago. In the 2000
MTF survey, 8th, 10th and 12th graders, respectively, reported binge drink-
ing rates of 14%, 26%, and 30%.[2] Available evidence also indicates that
alcohol use rates increase during the adolescent period, followed by a gradual
FAMILY EMPOWERMENT INTERVENTIONS 1361
decline as youths enter early adulthood;[1] and that the rates of alcohol use,
particularly problem drinking, are higher among males than females.
Rates of alcohol use and problem use of this drug are higher among
youths involved with the juvenile justice system than their counterparts in
the general population.[3] A number of sociodemographic and psychosocial
factors have been found related to alcohol use and problem use of this drug
among these youths. Alcohol use among juvenile oenders has been found
to be related to their experience of traumatic life events (e.g., witnessing a
shooting or killing outside their homes),[4] having parents with alcohol
or other drug abuse problems,[5] experiencing co-occurring mental health
problems,[68] friends use of alcohol and other drugs[9]reecting greater
acceptability of drug use in their environments,[10] and delinquent
behaviorincluding violence.[9,11,12] Further, these youths tend to have an
early onset of use, which has been found to increase the chances of developing
alcohol dependence later in life.[13]
Adolescent treatment engagement and retention remain key issues in
the eld. Treatment entry and retention present considerable challenges in
treating drug-abusing adolescents.[14] Many drug-involved youths do not
enter treatment or leave prematurely, with associated high rates of return
to drug use, involvement in crime, and increased risk of HIV/AIDS, other
health and social problems.[15] Although a number of retention strategies
have been found to be successful,[16] more work is needed.
Alcohol has also been found to be a signicant factor in adolescent
substance abuse treatment outcomes. For example, Jainchill et al.[17] found
that completing therapeutic community treatment and associating with
nondeviant peers increased the chances of better outcomes (e.g., reductions
in alcohol/other drug use and no involvement in crime). In another important
study, Brown et al.[18] found alcohol to be involved in relapse following
substance abuse treatment. Although only 1% of the youths reported alcohol
as their drug of choice upon entering the hospital-based treatment program,
alcohol was involved in 46% of initial post-treatment relapses. For these
youths, alcohol served as a gateway drug to the return to heavy drug
involvement. (Although nicotine has also been found to be a gateway drug,
its use was not addressed in this study.) Brown et al.[18] determined that
the use of alcohol by these youths posttreatment reected the perception
that, given the prevalence of its use among adolescents, alcohol presented
less serious problems than drugs further along the drug use sequence
(e.g., amphetamines, hallucinogens, cocaine). Drawing upon the ndings of
this study, youths use of alcohol represents an important marker of treatment
outcome in its own right, and as a risk factor in the use of other drugs.
The present paper presents results of analyses on the impact of a Family
Empowerment Intervention on youths alcohol use, implemented as part of
1362 DEMBO ET AL.
a National Institute on Drug Abuse funded study, the Youth Support Project
(YSP). The YSP was an experimental, prospective study designed to test the
eectiveness of an in-home service involving juvenile oenders and their
families. It was hypothesized that participation in the FEI would result
in reductions in their substance use, including alcohol, over a 36-month
follow-up period. Results indicate the signicant impact of this intervention
on the extent of use of alcohol among youths who completed the FEI.
METHOD
The Youth Support Project
As discussed in more detail elsewhere, the YSP implemented a 10-week

systems-oriented and structural approach to family preservation: a home-
based Family Empowerment Intervention.[19] Families involved in the pro-
ject were randomly assigned to one of two groups: the Extended Services
Intervention (ESI) or the Family Empowerment Intervention (FEI) group.
ESI group families received monthly phone calls to maintain contact with
project research assistants. Both FEI and ESI families had 24 hour a day,
seven days a week access to YSP sta, and to information on various
community resources via a project developed agency and services resource
le. YSP sta provided families in both groups with information about
dierent community agencies, and assisted them in obtaining appropriate
referrals to meet their needs. A detailed discussion on enrollment and
participation rates appears in Dembo and Schmeidler.[19]
FEI group families received personal in-home visits from project eld
consultants to work on the following goals:
. restore the family hierarchy;
. restructure boundaries between parents and children;
. encourage parents to take greater responsibility for family
functioning;
. increase family structure through implementation of rules and
consequences;
. enhance parenting skills;
. have parents set limits, expectations, and rules that increase the
likelihood the target youths behavior will improve;
. improve communication skills among all family members;
. improve problem-solving skills, particularly in the target youth; and
. where needed, connect the family to other systemssystem t
(e.g., school, church, community activities).
It was expected that this intervention would result in improvements in

the target youths behavior and psychosocial functioning. (The present paper
focuses on a specic, important aspect of youth behavior; information on
their satisfaction with their families and communication with parents is
discussed in Dembo and Schmeidler.[19]) FEI families were expected to parti-
cipate in three, one-hour family meetings per week for approximately 10 weeks;
this reected an understanding that 30 sessions are needed to produce
meaningful change.[20] All household members (i.e., persons living under the
same roof as the target youth) were expected to participate in these meetings.
The FEI evolved into a structured series of events with families, all
based upon its original design.[21] Our work with families enriched these
developments. Project implementation and activities manuals were developed
as complementary components of the intervention.[22] The implementation
manual provides the theoretical foundation for the FEI, together with the
policies and practices involved in carrying it out. The activities manual
presents specic games, artistic projects, and exercises in which family
members can engage to facilitate achieving the goals of the intervention.
The activities book evolved from our early experience in implementing the
FEI, which indicated many of the families we worked with did respond well to
verbal interaction and sharing of feelings and information. We found that the
various activities often provided an essential way to bring to light issues the
families were experiencing in a manner which they could directly understand.
These activities and exercises opened up important avenues of communica-
tion between eld consultants and their families, and facilitated the families
achievement of the goals of the intervention.
A distinctive feature of this intervention is that the families were served
by eld consultants, who were not trained therapistsalthough they were
trained by, and perform their work under the direction of, licensed clinicians
(eld consultant training is covered in Dembo and Schmeidler[19]). The choice
of paraprofessionals was based on a cost eectiveness argument, and is sup-
ported by experimental research indicating that, at least for some treatments,
paraprofessionals produce outcomes that are better than those under control
conditions, and similar to those involving professional therapists.[23,24]
Further, by requiring less previous therapy training, the FEI, if proven eec-
tive, is expected to be highly attractive to agencies providing services to
juvenile oenders, which often operate with nancial constraints.
Youths processed at the Hillsborough County Juvenile Assessment
Center[25] who were arrested on misdemeanor or felony charges were sampled
for inclusion in the project. When openings occurred on the eld consultants
case loads, a listing of recently arrested youths was drawn. A cross-tabulation
of these cases was completed in regard to their gender and race/ethnicity
(African-American, Latino, and Anglo); and equal numbers of youths in each
1364 DEMBO ET AL.
of these six cells were randomly selected to process for enrollment in the
Youth Support Project. This sampling procedure was designed to oversample
Hispanics and females. However, so few Hispanic females were available they
were not always sampled when a sample was selected for each other cell. The
sample included only 25 Hispanic females, 53 Hispanic males, and 58 to 60
youths in each of the other cells. Since the actual sample represented the
population of JAC youths better than had been intended, the sample was
not weighted to reproduce the demographic composition of the population.
A weighted analysis would have made the computations much more compli-
cated. Informed consent of the youths and family members to participate in
the project was obtained before the initial interviews of youths and families,
and randomization to the FEI or ESI.
Initial and Follow-Up Interviews
Initial interviews were completed with 315 youths processed at the

Hillsborough County Juvenile Assessment Center from September 1, 1994
through January 31, 1998. A detailed discussion of the interview questions/
instruments appears in Dembo and Schmeidler.[19] The data in this paper are
from the 278 (88%) youths who completed at least one follow-up interview.
Each youth was paid $10 for completing the one and one-half to two hour
initial interview. Depending on the year youths entered the project, up to
three annual follow-up interviews were sought for completion as follows:
. Period of initial interview (Year 1) . Follow-up interviews sought
September 1, 1994 to January 31, 1996 Year 2, Year 3, Year 4
February 1, 1996 to January 31, 1997 Year 2, Year 3
February 1, 1997 to January 31, 1998 Year 2
We achieved high year 2, year 3, and year 4 follow-up interview rates.
Completion rates were as follows: year 2: 88.2% of 315 sought interviews;
year 3: 85.0% of 200 sought interviews; year 4: 75.8% of 120 sought
interviews. The follow-up interviews averaged one and one-half hours. Each
youth was paid $20 for a completed follow-up interview. Importantly, we had
a low refusal rate for each follow-up interview. If we exclude youths who
moved out of state (who were not routinely followed-up) and youths
who could not be located, we achieved net reinterview success rates of
95.5%, 93.4%, and 91% for the year 2, year 3, and year 4 interviews, respect-
ively. There were no signicant dierences in the reinterview results for FEI
and ESI youths. At each reinterview wave, over 90% of the interviews were
completed within 120 days following the anniversary of the preceding
interview.
The data set used in the analyses consisted of year 2 interview data on
278 youths, year 3 interview data on 170 youths, and year 4 interview data on
91 youths. Appendix A presents the proportion of data present for the various
analyses we conducted. The youths were interviewed in a variety of settings at
each interview wave. A majority of youths were interviewed at home or in
another community location. However, other youths were interviewed in
Florida Department of Juvenile Justice residential commitment programs,
county jails, juvenile detention centers, or in another type of secure program
such as prison.
Comparison of the 278 Reinterviewed Youths with the

Other Youths in the Study
A discriminant analysis[26,27] was performed comparing the 278

reinterviewees with the other 37 youths in the study, who only had an initial
interview, to learn if there were any important dierences between the two
groups. The two groups were compared in regard to: their initial interview
demographic characteristics (age, gender, ethnicity, race, living situation);
family members alcohol or other drug use or mental health problems, and
family member contact with the justice system; lifetime self-reported history
of physical abuse, sexual victimization, and frequency of marijuana/hashish
use in blunt and nonblunt form, hallucinogen and cocaine use; lifetime
history of referrals to juvenile court on delinquency charges or for
dependency (i.e., physical abuse, sexual exploitation/victimization, neglect);
lifetime reported treatment for a substance abuse or mental health problem,
and the lag between the youths grade level and their chronological age;
friends substance use and involvement with the justice system; self-reported
delinquency in the year prior to initial interview and reported frequency of
getting drunk or very high on alcohol in the past year; RIAH hair test results
for marijuana, cocaine, and opiates; arrest charges upon entering the juvenile
assessment center (i.e., violence oenses, property oenses, drug oenses, or
public disorder misdemeanors); emotional/psychological functioning as
measured by the SCL-90-R;[28] being a diversion (minor oender) or more
serious oender case; and assignment to the ESI or FEI group.
The results indicated that, overall, the two groups were not signicantly
dierent from one another on these variables (chi-squared test of Wilks
lambda 44.59, df 32, p n.s.). The two groups were signicantly dierent
from one another on only three of the comparison variables: African-
American youths were more likely to be included among the reinterviews
(F 5.41, df 1,313, p .021); youths with positive hair test results for
marijuana were less likely to included in the reinterviews (F 4.15,
1366 DEMBO ET AL.
df 1,313, p .042); and youths charged with drug oenses upon entering
the assessment center were less likely to be included in the reinterviews
(F 4.74, df 1,313, p .030). However, none of these relationships were
signicant by the .05/32 .002 Bonferroni inequality criterion that takes
account of the number of predictors.
Demographic, Educational, and Treatment History

Description of the Youths
As Table 1 indicates, most youths were male (56%), and averaged 14.5
years of age. Fifty-six percent of the youths were Anglo; 41% were African-
American, and 26% were Latino. Seventeen percent of the youths indicated
they lived with both their biological parents. An additional 64% indicated
they resided with either their mother only (51%), mother and another adult
(5%), or mother and stepfather (8%). Information on the occupational
status of the household chief wage earner or other sources of household
income (derived from Fishburne, Abelson, and Cisin[29] ), a measure of
socioeconomic status, highlighted the low to moderate SES of the youths
families. Eleven percent of the chief wage earners held an executive,
administrative/managerial, or professional specialty type position; 13%
held technical, sales, or administrative support position; 5% held skilled
jobs; 37% held unskilled, semiskilled, or low/moderate skilled service
positions; and 8% of the youths households were supported by public
funds. Twenty-ve percent of the cases had missing or uncodable informa-
tion on this variable.
Although most of the youths (88%) were still attending school, sizable
proportions of them were experiencing educational problems. For example,
44% of the youths indicated they had been placed in a special educational
program (e.g., educationally handicapped, severe learning disorder); and
50% noted they had repeated a grade in school. Relatedly, most youths
(62%) lagged one or more grade levels behind the grade level that
would be expected based on their chronological age. Few youths reported
current (4%) or previous (16%) mental health treatment. Even fewer youths
reported current (1%) or previous (4%) substance misuse treatment.
Arrest Charges upon Entering JAC
Almost all the youths entered JAC as a result of being taken into
custody on one or more felony or misdemeanor charges. Many of the youths
were charged with felony property oenses (52%) (especially burglary,
Table 1. Demographic, Educational, and Treatment History Description of the

Youths in the Study
N %
Race
African-American 115 41%
Native American/Indian 1 <1%
Oriental 1 <1%
Anglo 156 56%
Other 5 2%
Total 278 99%
Age
11 13 5%
12 12 4%
13 45 16%
14 69 25%
15 60 22%
16 39 14%
17 39 14%
18 1 <1%
Total 278 100%
Mean 14.54
Standard deviation 1.60
Sex
Male 155 56%
Female 123 44%
Total 278 100%
Ethnicity
Latino 72 26%
Non-Latino 206 74%
Total 278 100%
Occupational status/income source of household head
Executive, administrative, 31 11%
and managerial occupations/
professional specialty
occupations
Technical, sales, 37 13%
administrative support
occupations
Skilled occupations 13 5%
Unskilled, semiskilled, 104 37%
and low or moderate
skilled service
occupations
(continued )
1368 DEMBO ET AL.
Table 1. Continued
N %
Public assistance/other 23 8%
public support
No information 70 25%
Total 278 99%
Living situation
Biological parents 47 17%
father 7 3%
Father and stepmother 6 2%
mother 142 51%
Mother and stepfather 21 8%
Mother and other adult 14 5%
Grandmother/grandfather 13 5%
Aunt 7 3%
Guardian 8 3%
Boyfriend 1 <1%
Other 11 4%
Total 276 101%
Repeated a grade
Yes 140 50%
No 138 50%
Total 278 100%
Currently in school
Yes 245 88%
No 33 12%
Total 278 100%
Placed in a special education class
Yes 123 44%
No 153 55%
No information 2 1%
Total 278 100%
Lag between grade and chronological age
2 grades above 2 1%
1 grade above 7 3%
At grades level 88 32%
1 grade below 122 44%
2 grades below 39 14%
3 grades below 11 4%
4 grades below 1 <1%
No information 8 3%
Total 278 101%
(continued )
Table 1. Continued
N %
Mental health treatment history

Never 219 79%
Previously 44 16%
Currently 11 4%
Refused to answer 2 1%
No information 2 1%
Total 278 101%
Substance abuse treatment history
Never 259 93%
Previously 12 4%
Currently 3 1%
Refused to answer 2 1%
No information 2 1%
Total 278 100%
grand larceny, or auto theft) or misdemeanor property charges (e.g., retail

theft) (41%). Relatively few youths were arrested on charges of felony (17%)
or misdemeanor violence (2%), on felony (6%) or misdemeanor drug charges
(4%), or on public disorder misdemeanors (8%).
Self-Reported Heavy Drinking
The interview questions regarding the youths alcohol use probed their
age of rst use, recency of use, the number of days used in the past month,
and the number of times the youth got very high or drunk on alcohol in the
past year. The analyses we report in this paper focused on the youths
responses to a question probing the number of times in the 12 months
prior to their initial interviews, and during each relevant follow-up period
(i.e., since their previous interview), they reported getting very high or drunk
on alcohol. The question was taken from the National Household
Survey on Drug Abuse.[30] As Table 2 indicates, reported getting very
high or drunk on alcohol was a coded frequency variable with the following,
approximately log scale, categories (other than 0): 0 not used, 1 1 or 2
days in the past 12 months, 2 3 to 5 days in the past 12 months, 3 every
other month or so (or 6 to 11 days a year), 4 1 to 2 times a month (or 12
to 24 times a year), 5 several times a month (or about 25 to 51 days a
1370 DEMBO ET AL.
Table 2. How Many Times Reported Getting Very High or Drunk on Alcohol
In Year During First During Second During Third

Before Initial Follow-Up Follow-Up Follow-Up
Frequency Interview Period Period Period
None 55% 55% 45% 46%

1 or 2 days 11% 11% 18% 16%
3 to 5 days 6% 6% 9% 12%
Every other month or
so (or 6 to 11 days) 9% 9% 5% 8%
1 to 2 times a month
(or 12 to 24 days a year) 6% 6% 8% 7%
Several times a month
(or 25 to 51 days a year) 3% 3% 6% 3%
About 1 or 2 days a week 5% 5% 6% 3%
Almost daily or
3 to 6 days a week 4% 4% 2% 2%
Daily 1% 1% 1% 2%
100% 100% 100% 99%
(N 278) (N 278) (N 170) (N 91)
year), 6 about 1 or 2 days a week, 7 almost daily (or 3 to 6 days a week),

and 8 daily.
As the results in Table 2 indicate, 19% of the youths reported they had
gotten very high or drunk on alcohol 12 or more days in the 12 months
preceding their initial interview. The youths reported similarly high rates of
getting very high or drunk on alcohol during their follow-up interviews.
Nineteen percent, 23%, and 17% of the youths reported getting very
high/drunk on alcohol during the rst, second, and third follow-up periods,
respectively (Table 2).
Although there is no biological test for long-term drinking, tests for
marijuana and cocaine provide contrasting conclusions on the validity of
self-reported use.[31] Most youths who tested positive for marijuana
acknowledged its use, but very few youths who tested positive for cocaine
reported recent use. Furthermore, a substantial proportion of youths who
tested negative for marijuana claimed marijuana use, but few who tested
negative for cocaine. These results were interpreted as suggesting that admit-
ting marijuana use is more socially acceptable than admitting cocaine use.
If so, it is likely that the youths described their alcohol use accurately, since it
is less stigmatizing than even marijuana.
RESULTS
Analytic Strategy
The growth model shown in Fig. 1 was estimated by AMOS (analysis

of moment structures) 4.0.[32,33] AMOS implements a general approach to
multivariate linear data modeling, causal modeling, and/or conrmatory
factor analysis. Lack of signicance of the chi-square test used to test the
t of the model to the data indicates acceptable model t. AMOS also
provides a number of t measures to assess the closeness of t of the model
to the data, including the root mean square error of approximation
(RMSEA),[34] indicating good t if it is low.
The time of entry into the intervention determined the number of
follow-up interviews each youth received. Hence, the alcohol use data
that are missing are a consequence of the design of the study. Accordingly,
the full-information maximum likelihood (FIML) feature of AMOS
was used to estimate model parameters when data are missing. When
AMOS encounters missing values in the data that are modeled, the
program switches from its default moment-based maximum likelihood
algorithm to the FIML procedure. FIML has been found to have
greater statistical eciency when compared to mean-substitution, listwise,
or pairwise deletion procedures.[35] While AMOSs FIML algorithm does
Figure 1. Linear growth curve model of the eect of the FEI on Youths reported
frequency of getting very high or drunk on alcohol.
1372 DEMBO ET AL.
not require imputation of missing data, it produces parameter estimates and

standard errors comparable to modern multiple imputation methods for
incomplete data.[3638] If desired, estimates of the missing values may be
obtained after model estimation, combining the estimated model parameters
with the observed portion of a data record.[35,39]
Consistent with our previous outcome analyses,[3,40] the model
shown in Fig. 1 was tted with treatment (TX) eects estimated for the
following group comparisons: (1) FEI vs. ESI group youths (IGROUP),
(2) FEI completers vs. all other cases (GRADGP), and (3) FEI completers
vs. FEI noncompleters (COMPLETE). In preparing for these analyses, a
discriminant analysis was completed comparing ESI youths, youths not
completing the FEI, and youths completing the FEI on their initial interview
characteristics (psychosocial, delinquency history, abuse-neglect history,
sociodemographic, and referral history variables). No signicant overall
dierences were found. In addition, there was no signicant dierence among
the three groups on any of the variables. Based on these results, it was not
deemed necessary to control statistically for possible dierences at entry
among these three groups before examining the impact of the intervention
on the youths reported frequency of getting very high or drunk on alcohol
over time. As Fig. 1 indicates, each linear growth model consisted of the
average (LEVEL), with coecient 1 at each interview, and a slope
(TREND), reecting the annual rate of change in reported frequency of
getting very high or drunk on alcohol, with coecients increasing with
timestarting with an implicit zero at Time 1 (baseline).
Test of the Model Comparing ESI and FEI Youths
Table 3 presents the results of our analysis of the t of the model

shown in Fig. 1 to the data involving IGROUP as the treatment (TX)
variable. Although the model ts the data (RMSEA .018) there is
no statistically signicant trend on IGROUP eect. The covariance
between trend and level is not statistically signicant (critical ratio 1.92,
p .055).
Test of the Model Comparing FEI Completers

and All Other Cases
As Table 4 shows, the model ts the data quite well. A statistically

signicant, negative relationship is found for trend on GRADGP, indicating
that the reported frequency of getting very high or drunk on alcohol declines
Table 3. Multivariate Linear Growth Curve Model of the Eect of the Family
Empowerment Intervention (FEI) on Youths Reported Frequency of Getting
Very High or Drunk on Alcohol: FEI vs. ESI Youths (N 278)
Maximum Likelihood Estimates Estimate S.E. Critical-Ratio
Regression weights
Trend on Igroup .025 .116 .216
Means:
Igroup .475 .030 15.833
Intercepts
Level 1.545 .124 12.460
Trend .039 .089 .438
Covariances
Other1 with Other2 .401 .209 1.919
Variances
Igroup .249 .021 11.857
Other1 2.384 .435 5.480
Other2 .418 .127 3.291
e1 2.450 .424 5.778
e2 2.547 .295 8.633
e3 2.122 .330 6.430
e4 1.017 .494 2.059
2 8.742, df 8, p .36, RMSEA .018, Pr(RMSEA<.05) .768.
over time more for FEI completers, compared to all other cases. Relatedly,
there is a statistically signicant covariance between Level with Trend,
indicating a reduction in the mean rate of frequency of getting very high
or drunk on alcohol over time.
Test of the Model Comparing FEI Completers

and FEI Noncompleters
The model ts the data quite well (Table 5). Further, there is a
statistically signicant, negative trend on COMPLETE eect, highlighting
that reported frequency of getting very high or drunk on alcohol
declines more over time for FEI completers than FEI noncompleters. The
covariance between Trend and Level is not statistically signicant (critical
ratio 1.87).
1374 DEMBO ET AL.
Very High or Drunk on Alcohol: FEI Completers vs. All Other Cases (N 278)
Regression weights
Trend on Gradgp .292 .130 2.246
Means
Gradgp .270 .027 10.000
Intercepts
Level 1.539 .124 12.411
Trend .119 .078 1.526
Covariances
Variances
Gradgp .197 .017 11.588
Other1 2.401 .436 5.507
Other2 .427 .128 3.336
e1 2.429 .424 5.729
e2 2.541 .294 8.643
e3 2.123 .328 6.472
e4 1.028 .493 2.085
Test of the Model Comparing FEI Completers vs.

All Other Cases and FEI Completers vs.
FEI Noncompleters Involving Demographic Covariates
Ideally, we would have tted the model across demographic sub-

groups. However, there were an insucient number of cases in the various
comparison groups to conduct these multigroup analyses. As an alternative
procedure, separate analyses were completed examining the t of the model
to the data for GRADGP and COMPLETE as intervention variables,
involving sex, age, race, and ethnicity as covariates.
Test of the Model with Demographic Covariates
We examined the t of the model shown in Fig. 2 to the data

involving GRADGP as the intervention variable. As the results shown
in Table 6 indicate, a statistically signicant negative relationship remains
Very High or Drunk on Alcohol: FEI Completers vs. FEI Noncompleters (N 132)
Regression weights
Trend on Complete .544 .160 3.400
Means
Complete .568 .043 13.209
Intercepts
Level 1.382 .176 7.852
Trend .410 .137 2.993
Covariances
Variances
Complete .245 .030 8.167
Other1 2.237 .583 3.837
Other2 .467 .184 2.539
e1 2.628 .596 4.408
e2 1.978 .350 5.651
e3 2.235 .474 4.715
e4 .957 .682 1.403
for Trend on GRADGP. In addition, statistically signicant eects exist

for Level on age and Level on race. According to these results, reported
frequency of getting very high or drunk on alcohol: (1) declines more over
time for FEI completers than all other cases, (2) increases with age, and (3)
is smaller among African-American youths, than Anglo youths. Sex, age,
ethnicity, and race are essentially unrelated to GRADGP and Trend.
A similar analysis was completed with COMPLETE as the interven-
tion variable. The results obtained were similar to those found for the t
of the model with demographic covariates involving GRADGP. Table 7
presents these results.
Test of the Model Involving Quadratic Growth
We also tested for a quadratic growth eect in the data by incorpo-

rating this factor in separate ttings of the model involving GRADGP and
1376 DEMBO ET AL.
Figure 2. Linear growth curve model of the eect of the FEI on youths reported
frequency of getting very high or drunk on alcohol with demographic covariates.
COMPLETE as the intervention eect. However, no statistically signicant

quadratic eects were found.
CONCLUSIONS
The results of the linear growth analyses highlighted that youths

completing the FEI reported signicantly fewer occasions of getting very
high/drunk on alcohol over time, than youths all other youths (Table 4), or
youths not completing the FEI (among youths assigned to receive FEI
services) (Table 5). Relatedly, there is a net change per year in the reported,
coded frequency of getting very high/drunk on alcohol of .17 for youths
completing the FEI, while all other youths reported a frequency increase of
.12 (or 9%) per year; and a net change per year in reported frequency of
getting very high/drunk on alcohol of .13 for youths completing the FEI,
while youths not completing the FEI reported an average increase of .41
Very High or Drunk on Alcohol: FEI Completers vs. All Other Cases (N 278)
with Demographic Covariates
Regression weights
Trend on Gradgp .294 .129 2.283
Level on Sex .293 .238 1.231
Level on Age .258 .074 3.497
Level on Ethnicity .450 .301 1.496
Level on Race .961 .268 3.585
Trend on Sex .131 .141 .929
Trend on Age .032 .044 .723
Trend on Ethnicity .022 .178 .122
Trend on Race .218 .159 1.370
Means
Sex .558 .030 18.683
Age 14.543 .096 151.353
Ethnicity .259 .026 9.839
Race .414 .030 13.980
Intercepts
Level 1.864 1.109 1.681
Trend .415 .658 .631
Covariances
Sex with Age .021 .048 .438
Sex with Ethnicity .021 .013 1.604
Sex with Race .018 .015 1.249
Age with Ethnicity .028 .042 .674
Age with Race .081 .048 1.699
Ethnicity with Race .096 .014 6.786
Race with Gradgp .014 .013 1.100
Ethnicity with Gradgp .006 .012 .485
Age with Gradgp .033 .043 .780
Sex with Gradgp .014 .013 1.034
Variances
Gradgp .197 .017 11.769
Other1 1.956 .403 4.851
Other2 .395 .126 3.139
Other3 .195 .017 11.470
(continued )
1378 DEMBO ET AL.
Table 6. Continued
e1 2.468 .414 5.966

e2 2.529 .287 8.811
e3 2.049 .320 6.405
e4 1.167 .500 2.335
2 15.641, df 16, p .48, RMSEA 0, Pr(RMSEA<.05) .923.
Very High or Drunk on Alcohol: FEI Completers vs. FEI Noncompleters
(N 132) with Demographic Covariates
Regression weights
Trend on Complete .542 .161 3.362
Level on Sex .012 .334 .035
Level on Age .335 .106 3.153
Level on Ethnicity .715 .428 1.672
Level on Race .749 .377 1.984
Trend on Sex .054 .204 .265
Trend on Age .061 .065 .935
Trend on Ethnicity .247 .262 .944
Trend on Race .233 .230 1.011
Means
Sex .538 .044 12.348
Age 14.712 .138 106.327
Ethnicity .235 .037 6.341
Race .379 .042 8.937
Intercepts
Level 3.087 1.618 1.908
Trend 1.123 .995 1.129
Covariances
Sex with Age .003 .069 .048
Sex with Ethnicity .018 .019 .951
Sex with Race .014 .021 .678
Age with Ethnicity .045 .059 .763
Age with Race .126 .068 1.850
(continued )
Table 7. Continued
Ethnicity with Race .081 .019 4.212

Race with Complete .011 .021 .508
Ethnicity with Complete .026 .018 1.388
Age with Complete .026 .069 .377
Sex with Complete .018 .022 .820
Variances
Complete .245 .030 8.093
Other1 1.801 .533 3.379
Other2 .443 .181 2.450
e1 2.566 .571 4.492
e2 1.988 .342 5.807
e3 2.230 .473 4.719
e4 1.033 .693 1.491
2 15.635, df 16, p .48, RMSEA 0, Pr(RMSEA<.05) .961.
(or 33%) per year. Our ndings conrm the signicant impact of FEI on
problem drinking behavior among youths completing the FEI. These results
conrm in a more impressive manner ndings previously obtained involving
multiple regression analyses of the alcohol use data.[3,19] (An additional
growth model analysis comparing FEI completers and ESI youths indicated
FEI completers reported lower frequencies of getting very high/drunk on
alcohol over time than ESI youths [critical ratio: 1.56; .10>p>.05].)
Latent growth analysis represents an exciting opportunity to assess the eec-
tiveness of prevention or intervention programs with longitudinal outcome
data. Growth models provide a number of advantages to evaluation research-
ers. Perhaps the most important capability is to test developmental theories
involving linear and nonlinear trajectories over time.[41]
Direct Cost Benets of the FEI
The FEI also can be expected to result in direct cost savings to

the juvenile justice system. We examined this issue by estimating the
direct cost savings to the juvenile justice system by providing FEI services
to diversion-eligible youths in Hillsborough County, Florida, where the
intervention was implemented and evaluated. This analysis incorporated
information on the costs of new arrests and incarceration of youths
1380 DEMBO ET AL.
receiving ESI and FEI services during each 12-month period following
random assignment. In our cost-benet analysis, we estimated the three-
year cumulative direct cost savings to the Hillsborough County juvenile
justice system by providing FEI services to the 3600 diversion-eligible
youths who are processed each year at the Juvenile Assessment Center
(JAC). Diversion (nonserious oender) cases were chosen because they
represented the least costly group of oenders to process and involve in
service programs. Justice system processing costs were provided by the fol-
lowing agencies: Hillsborough County Sheris Oce, States Attorneys
Oce, Public Defenders Oce, 13th Judicial Circuit Court
Administrators Oce, and the Florida Department of Juvenile Justice. In
regard to incarceration costs, from justice system records we determined the
number of days ESI and FEI youths were incarcerated during each cumu-
lative 12-month period following random assignment to the ESI or FEI
group. Some youths were incarcerated in county jail, juvenile justice
residential commitment programs, or Florida Department of Corrections
prisons, each of which has dierent daily costs. However, all incarcerated
youths spent time in juvenile detention facilities. For many youths juvenile
detention represented their only periods of incarceration. Hence, for illus-
trative purposes, we used the daily cost for placement in a juvenile detention
center ($95; communication from Florida Department of Juvenile Justice,
August 4, 2000) to calculate the average incarceration cost per category of
youths.
The results of our analysis indicate that, taking into account the cost
of providing FEI services, implementing the FEI for 3600 diversion cases
processed at the JAC would save the Hillsborough County juvenile justice
system $4.7 million in cumulative direct costs over a three-year period. Not
surprisingly, since the FEI is an overlay service, the year-1 costs involved in
providing FEI services exceed those of providing current 90-day diversion
services. By year 2, due primarily to lower incarceration rates for youths
provided FEI services, youths receiving FEI services are projected to save
the juvenile justice system $934,139. By the end of year 3, the cumulative
cost savings to the justice system as a result of providing FEI services rises to
$4.7 million.
The data are even more impressive when one considers that many
other, nonestimated indirect cost benets could also be expected to occur
as a result of providing FEI services, such as (1) improvements in the quality
of family life, (2) improvements in the behavior of other family members
toward one another and in the community, (3) reduced family use of
various public resources (e.g., welfare, health), (4) reduced productivity
losses due to alcohol/other drug abuse or mental health disorders,
(5) reduced productivity losses due to incarceration or involvement in
careers in crime, and (6) reduced victimization of others (see Dembo &
Schmeidler[19] for more details).
How the FEI Could Be Improved
At the same time, our experience indicated two key ways in which the
FEI could be improved: (1) booster sessions, and (2) continuing eorts at
enrollment and involvement.
Booster Sessions
Although our clinical trial of the FEI did not do so, we

strongly recommend that, wherever possible, booster sessions be made
available to families receiving FEI services. Such sessions could be made
available to families that have completed the intervention, as well as those
families that have stopped participating in the intervention for such reasons
as work schedules. These sessions, periodically provided for a brief time,
could be helpful in assisting graduated families to maintain the gains they
made when receiving FEI services, and to address any new issues they may
be facing. For families that have not completed the FEI, booster sessions
could address current issues they are facing, as well as provide an
opportunity to reengage in intervention services. The long-term impact of
the FEI was gratifying given that no booster sessions were provided to
families assigned to receive intervention services. We think it is highly
likely that even more impact would have been found if such sessions had
been available.[19]
Continuing Eorts at Family Involvement
Our experience indicated that involving families in the FEI was a

continuing need. Self-termination presented an ongoing challenge to the
eld consultants and clinical sta. Five major strategies were used to involve
families in the intervention:[19] (1) engaging (an interpersonal process
whereby the eld consultant connects with the family in an empathetic
and positive manner), (2) joining (another form of engagement with
the family; however, it typically denotes a process whereby the eld
consultant specically connects with one person or subsystem in the
family), (3) tracking (a strategy whereby the eld consultant identies sig-
nicant symbolic expression communicated by family members), (4) enact-
1382 DEMBO ET AL.
ment (a process in which family members are instructed to interact in their

typical manner and then shown how to modify their interactions in more
positive ways), (5) circular questions (a style of inquiry designed to reveal
patterns and connections; circular questioning invites family members to
reect on issues, explore individual perceptions, and address concerns in
a highly interactive manner), and (6) reframing (a technique of relabeling
a behavior by putting it into a new, more positive perspective).
Ensuring the Integrity of Intervention Services
The eective implementation of the FEI required ongoing eld con-

sultant training and related eorts to ensure its integrity. The maintenance
of quality services within the FEI included the following elements:
. Weekly meetings with clinical supervisors. Field consultants met
with their clinical supervisor once a week for 90 minutes to review
their families cases and to obtain guidance on strategies and
activities to use in helping families reach the interventions goals.
. Bi-weekly group supervision meetings. During these meetings,
eld consultants showed videotapes of family meetings to the
other eld consultants and clinical sta and obtained ideas and
coaching on how to work more eectively with family members.
. Weekly in-service training sessions. Each week, eld consultants
received 90 minutes of training on various topics relating to the
intervention, e.g., its theoretical foundations, the connection
between specic goals of the intervention and activities used in
working with families, new community agencies/services, and the
procedures to use for system ts.
. Bi-weekly clinical sta and project director meetings. At these meet-
ings, the project director and clinical sta (line clinical supervisors
and clinical coordinator) discussed administrative issues aecting
service delivery (e.g., coordination with other agencies supervising
client youths) and issues aecting the integrity of service delivery
(e.g., training topics); steps are taken to resolve these issues.
. Weekly project meetings. Each week, eld consultants, clinical sta,
and administrative sta (i.e., the project director and administrative
assistant/coordinator) met to discuss any issues aecting the
delivery of quality services (e.g., caseload size), share information
(e.g., new community services), review family enrollment into the
intervention, and rearm commitment to the FEI goals.
The FEI is a low-cost, innovative service for arrested youths and

their families. It is particularly useful for minority and inner-city youths
and their families, who have historically been underserved.[42,43] Many of
these families often lack the resources to access community-based services.
When seeking publicly-funded services, they are often required to visit
program oces at some distance from their homes. Culturally sensitive
sta, providing in-home services, can help eliminate barriers to treatment
and meet families on their own terms.
Juvenile justice systems throughout the United States are experienc-
ing increased workload and resource pressures. Juvenile crime and its
eects continue to increase, and there is growing awareness of the magni-
tude of these and related problems among various high-risk groups.[44]
The FEI is an informed, eective response to these issues. By seeking to
strengthen families, the intervention helps sustain the most basic of our
social institutions.
APPENDIX A
Proportion of Data Present for

Various Latent Growth Model Analyses
a. FEI Versus ESI Youths (IGROUP)(N 278):
ALCHIGH ALC2HIGH ALC3HIGH ALC4HIGH IGROUP

ALC1HIGH 1.000
ALC1HIGH .978 .978
ALC3HIGH .612 .590 .612
ALC4HIGH .327 .317 .324 .327
IGROUP 1.000 .978 .612 .327 1.000
b. FEI Completers Versus All Other Cases (GRADGP)(N 278):
ALCHIGH ALC2HIGH ALC3HIGH ALC4HIGH GRADGP

ALC1HIGH 1.000
ALC1HIGH .978 .978
ALC3HIGH .612 .590 .612
ALC4HIGH .327 .317 .324 .327
GRADGP 1.000 .978 .612 .327 1.000
1384 DEMBO ET AL.
c. FEI Completers Versus FEI Noncompleters (COMPLETE) (N 132):
ALCHIGH ALC2HIGH ALC3HIGH ALC4HIGH COMPLETE

ALC1HIGH 1.000
ALC1HIGH .985 .985
ALC3HIGH .614 .598 .614
ALC4HIGH .356 .348 .348 .356
COMPLETE 1.000 .985 .614 .356 1.000
Alcohol Variable Denitions
ALC1HIGH: Self-reported frequency of getting very high or drunk on

alcohol in the year before initial interview.
alcohol during rst follow-up period.
alcohol during second follow-up period.
alcohol during third follow-up period.
ACKNOWLEDGMENTS
The authors should like to thank clinical, intervention, and other

research sta for their contributions to this project. Great thanks are due
to project eld consultants for their work. We deeply appreciate the support
of Mr. Darrell Manning, supervisor of the Juvenile Assessment Center; he
was a great resource to our work. We deeply appreciate Ms. Marianne Bells
word processing this manuscript. Thanks are also due to Ms. Felicita Reid
for translating the abstract into Spanish, and to Dr. Sergei Paromchik for
translating the abstract into French.
REFERENCES
1. Substance Abuse and Mental Health Services Administration

(SAMHSA). Summary Findings from the 1999 National Household
Survey on Drug Abuse; SAMHSA: Rockville, MD, Web site: http://
samsha.gov/oas/NHSDAS/1999/TitlePage.htm. Accessed February
19, 2001.
2. Johnson, L.D.; OMalley, P.M.; Bachman, J.G. High School and

Youth Trends. NIDA INFOFAX. http://www.nida.nih.gov/Infofax/
HSYouthtrends.html. Accessed on February 19, 2001.
3. Dembo, R.; Seeberger, W.; Shemwell, M.; Schmeidler, J.; Klein, L.;
Rollie, M.; Pacheco, K.; Hartseld, A.; Wothke, W. Psychosocial
Functioning Among Juvenile Oenders 12 Months After Family
Empowerment Intervention. J. Oender Rehabil. 2000, 32, 156.
4. Crimmins, S.M.; Cleary, S.D.; Brownstein, H.H.; Spunt, B.J.; Warley,
R.M. Trauma, Drugs and Violence Among Juvenile Oenders.
J. Psychoactive Drugs 2000, 32, 4354.
5. Zhang, L.; Welte, J.W.; Wieczorek, W.F. The Inuence of Parental
Drinking and Closeness on Adolescent Drinking. J. Stud. Alcohol
1999, 60, 245251.
6. Clark, D.B.; Kirisci, L. Posttraumatic Stress Disorder, Depression,
Alcohol Use Disorders, and Quality of Life in Adolescents. Anxiety
1996, 2, 226233.
7. Winters, K.C. Substance Abuse and Juvenile Oenders; University of
Minnesota, Presentation at Physicians Leadership for National Drug
Policy Conference, Washington D.C., November 6, 1998.
8. Van den Bree, M.B.M.; Svikis, D.S.; Pickens, R.W. Genetic Inuences
in Antisocial Personality and Drug Use Disorders. Drug Alcohol
Depend. 1998, 49, 177187.
9. Dembo, R.; Williams, L.; Wothke, W.; Schmeidler, J. The
Relationships Among Family Problems, Friends Troubled Behavior
and High Risk Youths Alcohol/Other Drug Use and Delinquent
Behavior: A Longitudinal Study. Int. J. Addictions 1994, 29,
14191442.
10. Dembo, R. Drug Use Socialization and Maturation: Childhood and
Adolescence. In Encyclopedia of Criminology and Deviant Behavior;
Bryant, C.D., Ed.; Taylor & Francis: London, 2000.
11. Brounstein, P.; Hatry, H.; Altschuler, D.; Blair, L. Substance Use and
Delinquency Among Inner City Adolescent Males; Urban Institute
Press: Washington, DC, 1990.
12. Inciardi, J.A.; Horowitz, R.; Pottieger, A.E. Street Kids, Street Drugs,
Street Crime; Wadsworth: Belmont, CA, 1993.
13. Grant, B.F.; Dawson, D.A. Age at Onset of Alcohol Use and its
Association with DSM-IV Alcohol Abuse and Dependence: Results
from the National Longitudinal Alcohol Epidemiologic Survey.
J. of Substance Abuse 1997, 9, 103110.
14. Battjes, R.J.; Onken, L.S.; Delany, P.J. Drug Abuse Treatment Entry
and Engage: Report of a Meeting on Treatment Readiness. J. Clin.
Psychol. 1999, 55, 643657.
1386 DEMBO ET AL.
15. Brown, S.A. Recovery Patterns in Adolescent Substance Abuse. In

Addictive Behaviors Across the Life Span: Prevention, Treatment, and
Policy Issues; Marlatt, G.A., McMahon, R.J., Baer, J.S., Eds.; Sage:
Newbury Park, CA.
16. Winters, K.C. Treatment of Adolescents with Substance Use Disorders;
Center for Substance Abuse Treatment: Rockville, MD, 1999.
17. Jainchill, N.; Hawke, J.; DeLeon, G.; Yagelka, J. Adolescents in TCs:
One-year Post-treatment Outcomes. J. Psychoactive Drugs 2000, 32,
8194.
18. Brown, S.A.; Tapert, S.F.; Tate, S.R.; Abrantes, A.M. The Role of
Alcohol in Adolescent Relapse and Outcome. J. Psychoactive Drugs
2000, 32, 107115.
19. Dembo, R.; Schmeidler, J. Family Empowerment Intervention: An
Innovative Service for High-Risk Youths and Their Families; Haworth
Press: Binghamton, NY, in press.
20. Mikesell, R.H.; Lusterman, D.; McDaniel, S., Eds. Integrating
Family Therapy; American Psychological Association: Washington,
D.C., 1995.
21. Cervenka, K.A.; Dembo, R.; Brown, C.H. A Family Empowerment
Intervention for Families of Juvenile Oenders. Aggression Violent
Behav. 1996, 1, 205216.
22. Dembo, R.; Chase, J.; Lamson, K.; DuDell, G. Family Empowerment
Intervention Manual; University of South Florida, Department of
Criminology: Tampa, FL, 1999.
23. Christenson, A.; Jacobson, N.S. Who (Or What) Can Do
Psychotherapy: The Status and Challenge of Non-professional
Therapies. Psychol. Sci. 1994, 5, 814.
24. Weisz, J.R.; Weiss, B.; Han, S.S.; Granger, D.A.; Norton, B. Eects of
Psychotherapy with Children and Adolescents Revisited: A Meta-
analysis of Treatment Outcome Studies. Psychol. Bull. 1995, 117,
450468.
25. Dembo, R.; Brown, R. The Hillsborough County Juvenile Assessment
Center. J. Child Adolesc. Subst. Abuse 1994, 3, 2543.
26. Bennett, S.; Bowers, D. An Introduction to Multivariate Techniques
for Social and Behavioral Sciences; John Wiley & Sons: New York,
1976.
27. Klecka, W.R. Discriminant Analysis; Sage: Beverly Hills, CA, 1980.
28. Derogatis, L.D. SCL-90-R Administration, Scoring and Procedures
Manual; Clinical Psychometric Research: Towson, MD, 1983.
29. Fishburne, P.M.; Abelson, H.I.; Cisin, I. National Survey on Drug
Abuse: Main Findings1979; National Institute on Drug Abuse:
Rockville, MD, 1980.
30. National Institute on Drug Abuse. 1985 National Household Survey on

Drug Abuse Questionnaire; National Institute on Drug Abuse:
Rockville, MD, 1985.
31. Dembo, R.; Shemwell, M.; Guida, J.; Schmeidler, J.; Baumgartner, W.;
Ramirez-Garnica, G.; Seeberger, W. Comparison of Self-report, Urine
Sample, and Hair Testing For Drug Use: A Longitudinal Study. In
Drug Testing Methods: Assessment and Evaluation; Mieczkowski, T.,
Ed.; CRC Press: New York, 1999.
32. Arbuckle, J.L. Amos: Analysis of Moment Structures. Psychometrika
1994, 54, 135137.
33. Arbuckle, J.L.; Wothke, W. Amos 4.0 Users Guide; Smallwaters
Corporation: Chicago, IL, 1999.
34. Browne, M.W.; Cudeck, R. Alternative Ways of Assessing Model Fit.
In Testing Structural Equation Models; Bollen, K.A., Long, J.S., Eds.;
Sage: Newbury Park, CA, 1993.
35. Wothke, W. Longitudinal and Multi-group Modeling with Missing
Data. In Modeling Longitudinal and Multilevel Data: Practical Issues,
Applied Approaches and Specic Examples; Little, T.D., Schnabel, K.U.,
Baumert, J., Eds.; Lawrence Erlbaum Associates: Mahwah, NJ, 2000
(also available at http://www.smallwaters.com/whitepapers/).
36. Graham, J.W.; Hofer, S.M.; Donaldson, S.I.; MacKinnon, D.P.;
Schafer, J.L. Analysis with Missing Data in Prevention Research.
In The Science of Prevention: Methodological Advances from
Alcohol and Substance Abuse Research; Bryant, K., Windle, M.,
West, S., Eds.; American Psychological Association: Washington,
D.C., 1997.
37. Little, R.J.A.; Rubin, D.B. Statistical Analysis with Missing Data;
Wiley: New York, 1987.
38. Schafer, J.L. Analysis of Incomplete Multivariate Data; Chapman &
Hall: London, 1997.
39. Arbuckle, J.L. Full Information Estimation in the Presence of
Incomplete Data. In Advanced Structural Equation Modeling: Issues
and Techniques; Marcoulides, G.A., Schumacker, R.E., Eds.;
Lawrence Erlbaum Associates: Mahwah, NJ, 1996.
40. Dembo, R.; Ramirez-Garnica, G.; Rollie, M.; Schmeidler, J.;
Livingston, S.; Hartseld, A. Youth Recidivism 12 Months After a
Family Empowerment Intervention: Final Report. J. Oender
Rehabil. 2000, 31, 2965.
41. Curran, P.J.; Muthen, B.O. Testing Developmental Theories in Inter-
vention Research: Latent Growth Analysis and Power Estimation;
University of California, Graduate School of Education &
Information Studies: Los Angeles, 1996.
1388 DEMBO ET AL.
42. Arcia, E.; Keyes, L.; Gallagher, J.J.; Herrick, H. National Portrait of
Sociodemographic Factors Associated With Underutilization of
Services: Relevance to Early Intervention. J. Early Intervention 1993,
17, 283297.
43. Dembo, R.; Seeberger, W. The Need for Innovative Approaches to Meet
the Substance Abuse and Mental Health Service Needs of Inner-city,
African-American Youth Involved with the Juvenile Justice System;
Invited paper presented to the U.S. Commission on Civil Rights,
Washington, D.C., April, 1999.
44. Butts, J.A.; Harrell, A.V. Delinquents or Criminals: Policy Options for
Young Oenders. Washington, D.C.: Urban League, 1998.
RESUMEN
Le informamos los resultados obtenidos en el estudio analiticu hecho

sobre el impacto en la Familia del modelo de Intervencion llamado
Intervencion para el Fortalecimiento de la Familia (FEI). Los partici-
pantes de este estudio fueron jovenes que beb an en exeso. Este estudio
tuvo un per odo de seguimiento de 36 meses. Los jovenes que participaron
lo hicieron en el Centro de Evaluacion Juvenil del Condado de
Hillsborough. Las familias que participaron fueron seleccionadas al azar
con el n de determinar cual de dos lipos de intervenciones hechas era
estadisticamente signicativa. Un grupo de familias uso el tipo de inter-
vencion llamado Intervencion de Servicios Elitendidos (ESI) y el otro
recibio FEI. Las familias que les toco la forma de intervencion ESI reci-
bieron una llamada telefonica mensual con el proposito prestar apoyo en
la forma de informacion con los referidos apropiados de acuerdu con la
necesidad de la familia. Las familias que les toco la forma de intervencion
FEI recibieron en la casa de cada familia participante tres visitas a la
semana de una hora de duracion cada visita. Estas visitas fueron hechas
en un per odo de diez semanas. Las visitas las hizo una persona entrenada
al nivel para-profesional en el area de servicios cl nicos. La hipotesis a
probarse es que por medio de mejorar el funcionamiento de la familia
habilitando a los padres en su funcion se pudiera mejorar el comporta-
miento y el funcionamiento psicosocial de los jovenes selecionados para
participar en el estudio. Aunque la diferencia entre los metodos FEI y ESI
no fue estad sticamente signicativa, la frecuencia documentada de la posi-
bilidad de hallar a un participante embriagado o endrogado bajo mas
dentro del transcurso de tiempo seleccionado para el grupo trabajando
con la intervencion FEI para las familias que completaron el estudio
comparadas con las que no lo hicieron. El resultado obtenido apoya la

premisa de que el metodo de intervencion FEI inuye en el bienestar de las
familias que participan esa forma de intervencion.
RE SUME
Les resultats de notre analyse de mode`le de croissance de Iimpact de

Family Empowerment Intervention (FEI) pour les personnes intoxiquees
dalcool sont communiques. Ils sont obtenus pendant une periode de 36
mois pour les jeunes enregistres au centre devaluation de Hillsborough
County Juvenile. Les familles impliquees dans le projet ont ete aleatoirement
aectees pour recevoir soit Extended Services Intervention (ESI) soit la FEI.
Les familles aectees a` lESI ont recgu des appels telephoniques mensuels et,
si cest indique, les informations de reference. Les familles aectees a` la FEI
ont recu trois visites dune heure hebdomadaires au domicile pendant une
periode de 10 semaines environ dun para-professionnel forme dans une
clinique. En essayant dameliorer le fonctionnement de famille par la voie
dautorisation des parents, il a ete pose une hypothe`se que le comportement
de jeunes et leur fonction psychosociale amelioraient. Bien que la dierence
entre la FEI et lESI netant pas importante, la frequence de consommation
excessive dalcool ou detat ivre est declinee avec le temps plus dans le
groupe de la FEI que dans le groupe qui netait pas aecte a` la FEI. Les
resultats obtenus prouvent le succe`s des services de la FEI.
THE AUTHORS
Richard Dembo, Ph.D., is a Professor of

Criminology at the University of South
Florida in Tampa. He has a long-term
interest in developing, implementing, and
evaluating intervention programs for
high-risk youth.
1390 DEMBO ET AL.
Werner Wothke, Ph.D., is President of

SmallWaters Corporation, which pub-
lishes and supports multivariate software
programs, including AMOS (Analysis of
Moment Structures). He is an expert in
multivariate analysis procedures and in
psychosocial measurement.
Stephen Livingston, B.A., is a research

assistant in the Department of
Criminology at the University of South
Florida. He has been associated with the
Youth Support Project since 1998.
James Schmeidler, Ph.D., is an Assistant

Clinical Professor in the Departments of
Psychiatry and Biomathematical Sciences
at the Mt. Sinai School of Medicine. He
has considerable experience applying
statistical procedures to behavioral
science data.

The Impact of A Family Empowerment Intervention On Juvenile Offender Heavy Drinking

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Impact of A Family Empowerment Intervention On Juvenile Offender Heavy Drinking

Uploaded by

Copyright:

Available Formats

MARCEL DEKKER, INC.

270 MADISON AVENUE NEW YORK, NY 10016

SUBSTANCE USE & MISUSE

THE IMPACT OF A FAMILY

Richard Dembo, Ph.D.,1,y Werner Wothke, Ph.D.,2

We report the results of a growth model analysis of the impact

*Preparation of this manuscript was supported by Grant # 1-R01-DA08707, funded

DOI: 10.1081/JA-120014082 1082-6084 (Print); 1532-2491 (Online)

1360 DEMBO ET AL.

processed at the Hillsborough County Juvenile Assessment

Key Words: Family empowerment intervention; Alcohol

By any criterion, alcohol use among U.S. youths remains at an

FAMILY EMPOWERMENT INTERVENTIONS 1361

1362 DEMBO ET AL.

The Youth Support Project

As discussed in more detail elsewhere, the YSP implemented a 10-week

FAMILY EMPOWERMENT INTERVENTIONS 1363

It was expected that this intervention would result in improvements in

1364 DEMBO ET AL.

Initial and Follow-Up Interviews

Initial interviews were completed with 315 youths processed at the

FAMILY EMPOWERMENT INTERVENTIONS 1365

Comparison of the 278 Reinterviewed Youths with the

A discriminant analysis[26,27] was performed comparing the 278

1366 DEMBO ET AL.

Demographic, Educational, and Treatment History

Arrest Charges upon Entering JAC

FAMILY EMPOWERMENT INTERVENTIONS 1367

Table 1. Demographic, Educational, and Treatment History Description of the

1368 DEMBO ET AL.

FAMILY EMPOWERMENT INTERVENTIONS 1369

Mental health treatment history

grand larceny, or auto theft) or misdemeanor property charges (e.g., retail

Self-Reported Heavy Drinking

1370 DEMBO ET AL.

In Year During First During Second During Third

None 55% 55% 45% 46%

year), 6 about 1 or 2 days a week, 7 almost daily (or 3 to 6 days a week),

FAMILY EMPOWERMENT INTERVENTIONS 1371

The growth model shown in Fig. 1 was estimated by AMOS (analysis

1372 DEMBO ET AL.

not require imputation of missing data, it produces parameter estimates and

Test of the Model Comparing ESI and FEI Youths

Table 3 presents the results of our analysis of the t of the model

Test of the Model Comparing FEI Completers

As Table 4 shows, the model ts the data quite well. A statistically

FAMILY EMPOWERMENT INTERVENTIONS 1373

Maximum Likelihood Estimates Estimate S.E. Critical-Ratio

2 8.742, df 8, p .36, RMSEA .018, Pr(RMSEA<.05) .768.

Test of the Model Comparing FEI Completers

1374 DEMBO ET AL.

Maximum Likelihood Estimates Estimate S.E. Critical-Ratio

2 7.472, df 8, p .49, RMSEA .000, Pr(RMSEA<.05) .844.

Test of the Model Comparing FEI Completers vs.

Ideally, we would have tted the model across demographic sub-

Test of the Model with Demographic Covariates

We examined the t of the model shown in Fig. 2 to the data

FAMILY EMPOWERMENT INTERVENTIONS 1375

Maximum Likelihood Estimates Estimate S.E. Critical-Ratio

2 9.212, df 8, p .32, RMSEA .034, Pr(RMSEA<.05) .550.

for Trend on GRADGP. In addition, statistically signicant eects exist

Test of the Model Involving Quadratic Growth

We also tested for a quadratic growth eect in the data by incorpo-

2 8.742, df 8, p .36, RMSEA .018, Pr(RMSEA<.05) .768.

2 7.472, df 8, p .49, RMSEA .000, Pr(RMSEA<.05) .844.

2 9.212, df 8, p .32, RMSEA .034, Pr(RMSEA<.05) .550.

2 15.641, df 16, p .48, RMSEA 0, Pr(RMSEA<.05) .923.

2 15.635, df 16, p .48, RMSEA 0, Pr(RMSEA<.05) .961.