Juvenile Justice: Minimal Gender Bias Occurred in Processing Noncriminal
Juveniles (Letter Report, 02/28/95, GAO/GGD-95-56).
This report studies gender bias in state juvenile justice systems'
handling of status offenders, who are youths who have committed an
offense, such as truancy or ungovernable behavior, that would not be a
crime if committed by an adult. GAO defines "gender bias" as
intentional or unintentional differences in the juvenile justice
system's outcomes of female and male status offenders who have similar
characteristics, such as age, status offense, and prior offense history.
GAO (1) compares the outcome of the intake decisions and the frequency
and outcomes of detentions, adjudications, and out-of-home placements of
female and male status offenders and (2) compares the availability of
facilities and services for female and male status offenders in selected
jurisdictions.
--------------------------- Indexing Terms -----------------------------
REPORTNUM: GGD-95-56
TITLE: Juvenile Justice: Minimal Gender Bias Occurred in
Processing Noncriminal Juveniles
DATE: 02/28/95
SUBJECT: Juvenile status offenders
Juvenile delinquency
Correctional facilities
Detention facilities
State programs
Federal aid for criminal justice
Demographic data
Comparative analysis
Women
Discrimination
IDENTIFIER: Arizona
California
Florida
Missouri
Nebraska
South Carolina
Utah
Duval County (FL)
Fayette County (KY)
Johnson County (KY)
Bexar County (TX)
Alachua County (FL)
Prince George's County (MD)
St. Mary's County (MD)
Dallas (TX)
Jefferson County (KY)
******************************************************************
** This file contains an ASCII representation of the text of a **
** GAO report. Delineations within the text indicating chapter **
** titles, headings, and bullets are preserved. Major **
** divisions and subdivisions of the text, such as Chapters, **
** Sections, and Appendixes, are identified by double and **
** single lines. The numbers on the right end of these lines **
** indicate the position of each of the subsections in the **
** document outline. These numbers do NOT correspond with the **
** page numbers of the printed product. **
** **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced. Tables are included, but **
** may not resemble those in the printed version. **
** **
** Please see the PDF (Portable Document Format) file, when **
** available, for a complete electronic file of the printed **
** document's contents. **
** **
** A printed copy of this report may be obtained from the GAO **
** Document Distribution Center. For further details, please **
** send an e-mail message to: **
** **
** **
** **
** with the message 'info' in the body. **
******************************************************************
Cover
================================================================ COVER
Report to Congressional Requesters
February 1995
JUVENILE JUSTICE - MINIMAL GENDER
BIAS OCCURRED IN PROCESSING
NONCRIMINAL JUVENILES
GAO/GGD-95-56
Juvenile Gender Bias
(185010)
Abbreviations
=============================================================== ABBREV
NCJJ - National Center for Juvenile Justice
OJJDP - Department of Justice's Office of Juvenile Justice and
Delinquency Prevention
Letter
=============================================================== LETTER
B-259793
February 28, 1995
The Honorable Orrin G. Hatch,
Chairman
The Honorable Joseph R. Biden, Jr.,
Ranking Minority Member
Committee on the Judiciary
United States Senate
The Honorable William F. Goodling,
Chairman
The Honorable William L. Clay,
Ranking Minority Member
Committee on Economic and
Educational Opportunity
House of Representatives
The Juvenile Justice and Delinquency Prevention Amendments of 1992
(P.L. 102-586) requires us to study gender-bias issues in state
juvenile justice systems' handling of status offenders. Youth who
have come in contact with the juvenile justice system by committing
an offense (such as liquor offenses, running away, truancy, or
ungovernable behavior) that would not be a crime if committed by an
adult are status offenders.\1 For purposes of this review, we have
defined "gender bias" as intentional or unintentional differences in
the juvenile justice systems' outcomes of female and male status
offenders who (except for gender) have similar characteristics, such
as age, status offense, and prior offense history. As agreed with
your Committees, our specific objectives were to (1) compare the
outcome of the intake\2 decisions and the frequency and outcomes of
detentions, adjudications, and out-of-home placements of female and
male status offenders\3 and (2) compare the availability of
facilities and services for female and male status offenders in
selected jurisdictions.
--------------------
\1 Delinquents are juveniles who are found to have committed an
offense that would be criminal if committed by an adult.
\2 Intake is the process during which a juvenile referral or
complaint is received by the court or prosecutor's office.
\3 Detention is the placement of youth who may have committed a
status offense in a facility, which may be secure, while awaiting
processing by juvenile justice authorities. The cases of those
youths who are formally processed may involve adjudication and
disposition hearings. At an adjudication hearing, the juvenile court
determines if the youth has committed a status offense. If so, at a
concurrent or a subsequent disposition hearing, the judge determines
an appropriate action or treatment plan for the status offender. The
juvenile court judge's disposition options include dismissal of the
case, probation, fine or restitution, community service, and
placement in an out-of-home facility.
RESULTS IN BRIEF
------------------------------------------------------------ Letter :1
On the basis of our models using National Center for Juvenile Justice
(NCJJ)\4 data, the county probation officers' responses to our
survey, and our visits to selected facilities, we found minimal
gender bias, as we defined it.
According to NCJJ's national data, 500,620 status-offender cases were
petitioned to juvenile courts in the United States during the 6-year
period from 1986 to 1991. (These data did not include youth who were
handled informally--picked up, counseled, and/or released.) Of these
cases, about 40 percent involved females and 60 percent involved
males. Our analyses showed that females and males had similar
probabilities (percent chances) of being detained, adjudicated, or
placed for about 60 percent of the offense categories. For the
remaining 40 percent, females and males had different probabilities
for the 3 outcomes. On the basis of the relative numbers of cases
involving females versus males, female runaway cases outnumbered male
runaway cases by about 1-1/2 times, while male liquor offender cases
outnumbered female liquor offender cases by about 3 times. However,
our national-level analyses alone cannot be used to draw conclusions
about either the presence or the absence of gender bias. Such
conclusions cannot be drawn because NCJJ's national data files did
not contain sufficient information on prior histories of status
offenders and other variables relevant to judges' decisions in the
cases. Therefore, to enable us to determine if gender bias existed,
we developed 25 statistical models using data from those states with
appropriate data for these more in-depth analyses.
In comparing outcomes for female and male status offenders in
relationship to gender bias, each of the 25 models considered and
accounted for case characteristics--such as each youth's prior
offense history and age--that, according to the models, generally
influence intake and judicial decisions. For five of the six intake
regression models, our results indicated no evidence of gender bias.
Similarly, for 14 of the 19 regression models for the detention,
adjudication, and placement decisions, our results indicated no
evidence of gender bias in the juvenile courts' handling of status
offenders. However, for the one intake model that exhibited a
difference for a specific state, females were more likely to be
petitioned to juvenile court than males. For the other five
state-specific models--three detention, one adjudication, and one
placement--females were less likely to be detained, adjudicated, or
placed than males. Our analysis also showed that certain factors,
such as offenders' prior offense history and source of referral,
affected the status offenders' outcomes. While our models indicated
no evidence of gender bias in 19 cases and some evidence in 6 cases,
our results were limited to those aspects of the juvenile justice
system for which data existed.
At the 15 facilities we visited, we generally found minimal
gender-based differences in the availability of counseling,
educational, and medical services for females and males, although the
extent of such services varied by type of facility. The 10
co-educational facilities offered similar services for females and
males, although some of the facilities provided females with health
services, such as gynecological services and prenatal care, not
applicable to males. Officials at the other five facilities (four
serving females and one serving males) said that their programs were
not gender-based and could be replicated or offered to either females
or males. The only gender-based difference we noted involved
admission physicals. At two of the female-only group homes, health
examinations included testing (which could be refused at one of the
homes) for sexually transmitted diseases, whereas, at similar
male-only facilities operated by the same organizations, such testing
was not done unless requested by the males.
The survey responses from the county probation officers did not
indicate any significant gender-bias concerns. The respondents
reported that generally no differences existed in the way females and
males with similar status-offense histories were treated within their
juvenile justice systems.
About 70 percent of juvenile probation officers who responded for
their departments reported that treatment options, e.g., facilities
and services, were about equally available for detained female and
male status offenders. While some respondents noted that overall
more facilities were needed for both females and males, they added
that there were more facilities for males than females.\5
Officials in the jurisdictions and at the facilities we visited
emphasized that, in their view, insufficient facilities and services
were available for status offenders, irrespective of gender. These
officials said that more resources were needed for female and male
status offenders. These needed resources included early-intervention
programs to divert first-time offenders from further involvement in
the juvenile justice system. Officials said that status offenders
did not need gender-specific services, except for gynecological
services and prenatal care. The officials had mixed views about
whether the needs of status offenders were better met by
co-educational or single-gender facilities.
--------------------
\4 NCJJ, which is located in Pittsburgh, PA, is the research division
of the National Council of Juvenile and Family Court Judges.
\5 The respondents did not give us data on the number of males or
females who could have been placed in the facilities. Therefore, we
could not determine if disproportionately more facilities exist for
males than females.
BACKGROUND
------------------------------------------------------------ Letter :2
Juvenile justice is primarily the domain of state and local
authorities. Thus, juvenile courts' jurisdiction and procedures can
vary widely throughout the United States. For instance, depending
upon the state and the alleged offense, the juvenile courts'
jurisdiction may end at age 18, 17, 16, or even younger. Referrals
of youth to juvenile justice authorities can come from various
sources, including police officers, parents, schools, and social
service agencies. Police officers account for 41 percent of the
referrals, according to 1989 Department of Justice's Office of
Juvenile Justice and Delinquency Prevention (OJJDP) data.
Generally, after an alleged status offender is referred to juvenile
authorities, screening or intake staff (e.g., a juvenile probation
officer) decide whether the case should be handled formally or
informally. Juveniles can be temporarily placed in detention centers
at some point between referral and case disposition by the court. If
the intake decision were to proceed formally, a petition is drafted
and filed to provide notice of the offenses that will be pursued.
The petition charges the youth with a status-offense violation and
identifies the youth and those other persons who should be informed
of the proceedings. These proceedings include an adjudication
hearing and possibly a disposition hearing. At the adjudication
hearing, the juvenile court judge reviews evidence and determines if
the youth has committed a status offense. At a concurrent or
subsequent disposition hearing, the judge determines an appropriate
action or treatment plan for the status offender. The juvenile court
judge's disposition options include dismissal of the case, probation,
fine or restitution, community service, and placement. Placement
refers to any "out-of-home" disposition, which usually takes place in
residential facilities. These facilities provide 24-hour care to
juveniles. The following are types of residential facilities:
-- Detention centers: secure, residential facilities.
-- Group homes: nonsecure facilities that are intended to provide
a residential environment in which to meet the long-term
counseling needs of troubled youth.
-- Shelters: nonsecure facilities that are intended to provide
overnight or short-term housing and crisis intervention
counseling to troubled youth.
Figure 1 shows the number of petitioned juvenile cases processed by
the juvenile courts in 1991, according to OJJDP data.
Figure 1: Juvenile Court
Processing of Petitioned
Status-Offense Cases, 1991
(See figure in printed
edition.)
Note: Percentages may not add to 100 percent due to rounding.
Source: OJJDP data.
Cases handled informally usually do not involve a petition or an
adjudication hearing. These informal (nonpetitioned) cases may be
dismissed; possible reasons for dismissal include lack of evidence or
the youth's receiving a warning or counseling. Even when cases can
be handled informally, juveniles can be given probation or even
placed. As shown in figure 1, 2 percent (or 500) of all petitioned
nonadjudicated cases in 1991 resulted in juveniles' being placed.
According to OJJDP, in many jurisdictions, most status-offense cases
are handled informally. In many communities, county attorneys,
family crisis units, or social service agencies--rather than the
juvenile courts--have assumed responsibility for screening and
diverting alleged status offenders from the juvenile justice system.
Even though juvenile justice is primarily the responsibility of state
and local authorities, Congress has taken an increased interest in
juvenile justice issues during the past two decades. Most
significantly, the Juvenile Justice and Delinquency Prevention Act of
1974, as amended (42 U.S.C. 5601 et seq.), established a formula
grant program for states to improve their juvenile justice systems.
States receive formula grant funds if they comply with certain
requirements. One of these requirements was that status offenders
should not be held in secure detention facilities, such as jails,
police lockups, juvenile detention centers, or training schools. In
1980, Congress amended the law to allow states to detain status
offenders under certain conditions and still receive their grant
funds. According to OJJDP regulations, these status offenders must
be provided certain procedural protections.\6
Some child advocacy groups have raised concerns about the lack of
appropriate placement services for females in the juvenile justice
system. For example, in September 1992, the National Network for
Runaway and Homeless Youth Services advocated the need to review
gender bias within the states' juvenile justice systems. In
addition, some studies have indicated that females were more likely
to be detained for status offenses than males.
--------------------
\6 In 1991, we reported on the states' use of this amendment. See
our report entitled Noncriminal Juveniles: Detentions Have Been
Reduced but Better Monitoring Is Needed (GAO/GGD-91-65, Apr. 24,
1991).
SCOPE AND METHODOLOGY
------------------------------------------------------------ Letter :3
To aid us in defining gender bias and in designing models or
approaches to address the objectives, we reviewed relevant literature
identified in bibliographies provided us by NCJJ and OJJDP.
Regarding the first objective, we used NCJJ's national estimates of
status-offender data for calendar years 1986 through 1991 to develop
gender-specific probabilities of detentions, adjudications, and
placements for status offenders by offense categories. However,
these data did not contain sufficient information relevant to judges'
decisions to assess gender bias, e.g., prior offense history and
source of referral for the offense.\7 To examine gender bias, we did
further analysis using data from several states that had additional
variables beyond those used for NCJJ's national estimates. We
developed 6 models to study the outcome of intake decisions in 6
states and 19 models to further study detentions, adjudications, and
placements in 7 states. We used a class of models commonly used in
criminological research to analyze these types of outcomes. We used
NCJJ's state-specific data files to conduct regression analyses\8 for
seven states--Arizona, California, Florida, Missouri, Nebraska, South
Carolina, and Utah. Data limitations precluded us from developing
models for status offender intake decisions in Nebraska, placements
in Arizona, and detentions in Utah. Further, we could not address
possible gender bias elsewhere in the juvenile justice system because
data did not exist. For example, the data did not include youths who
were handled informally--picked up, counseled, and/or released by the
police or by county juvenile department intake officials.
To compare the availability of facilities and services, we visited a
total of 15 facilities located in 9 counties--generally 2 counties (a
rural county and an urban county) within each of 4 selected states
(Florida, Kentucky, Maryland, and Texas).\9
We mailed a survey to a national sample of county probation
department officials to obtain (1) opinions on differences in the
juvenile justice systems' processing of status offenders and (2)
perspectives on the availability of facilities and services for
status offenders. By using a national sample, we were able to
project the results to our study population of 1,249 chief juvenile
justice probation officers.
Appendix I presents more details about our objectives, scope, and
methodology, including a discussion of how we selected states for
analysis with respect to our second objective. Appendix V contains a
copy of the survey and the survey's results.
We did our work from March 1993 through August 1994 in accordance
with generally accepted government auditing standards. Since no
federal agency has responsibility for the issues discussed in this
report, we did not obtain official comments on a draft of this
report. However, we did discuss our results with NCJJ and OJJDP
officials and, where appropriate, incorporated their comments.
--------------------
\7 At the time of our review, NCJJ did not have national statistics
available for calendar year 1992. The national NCJJ data did not
represent the universe of status offenders. Rather, the data
included only those status offenders who were petitioned to or
otherwise handled more formally by the juvenile courts.
\8 Regression analysis was used to identify causal relationships
between or among two or more key variables, such as gender and age of
the juvenile, current offense category, and prior offense categories
and dispositions.
\9 We visited three counties in Kentucky (see table 5).
MINIMAL GENDER BIAS IN JUVENILE
JUSTICE SYSTEMS' HANDLING OF
STATUS OFFENDERS
------------------------------------------------------------ Letter :4
Our analyses of 6 years of national data indicated that there were
only relatively small differences in the percentages of female and
male status offenders detained, adjudicated, and placed. With six
exceptions, our logistic regression analyses\10 of intake decisions,
detentions, adjudications, and placements in seven states generally
did not indicate any significant gender-based differences in the
processing of female and male status offenders. In addition, our
national survey of county probation officers and site visits did not
identify any specific gender differences in juvenile justice systems.
--------------------
\10 Logistic regression analysis is a widely accepted statistical
methodology used when the dependent variable is qualitative, such as
if a status offender is detained. Regression analysis identifies
relationships between the dependent variable and two or more key
variables, such as gender and age of the juvenile, offender's current
offense category, and prior offense categories and dispositions.
NATIONAL DATA REVEALED SMALL
PERCENTAGE DIFFERENCES
---------------------------------------------------------- Letter :4.1
According to NCJJ national data, a total of 500,620 status- offender
cases were petitioned to juvenile courts during calendar years 1986
through 1991.\11 Of the total petitioned status-offender cases, 41.3
percent (206,756 cases) involved females and 58.7 percent (293,864
cases) involved males. In terms of gender distinctions, two specific
offense categories had noticeable differences in the numbers:
females were involved in 61.9 percent of the running away offenses
and males were involved in 74.3 percent of the liquor offenses.
Table 1 shows that petitioned female status offenders had about the
same probability, or percent chance, as petitioned male status
offenders of being detained, adjudicated, or placed out-of-home
during 1986 through 1991, for 60 percent of the outcomes. For
example, the probabilities for female and male truants who were
detained, adjudicated, or placed were within 2 percentage points of
each other. The exceptions were in the offense categories of running
away and liquor violations. For the offense categories of liquor
violation, running away, truancy, and ungovernability, our data
analysis showed that the probabilities of either female or male
status offenders' being detained before disposition by the juvenile
courts had declined from calendar years 1986 to 1991. For example,
the probability of ungovernable female status offenders' being
detained decreased from about 19 percent to 8 percent, and the
decrease for males was from 19 percent to 9 percent (see app. II,
table II.3).
Table 1
Comparative Probabilities (by Status-
Offense Category and Gender) Regarding
Detentions, Adjudications, and
Placements for Calendar Years 1986-1991
Adjudicated
status
offenders
Status offense (category and Detain Adjudicate receiving
gender) ed\a d\b placements\c
---------------------------------- ------ ---------- --------------
Liquor offense
----------------------------------------------------------------------
Female 4.96 57.34 5.68
Male 6.37 59.69 8.42
Running away
----------------------------------------------------------------------
Female 22.54 49.04 31.25
Male 28.28 52.49 34.68
Truancy
----------------------------------------------------------------------
Female 3.31 67.51 9.17
Male 3.90 69.51 10.35
Ungovernable
----------------------------------------------------------------------
Female 13.61 65.97 32.11
Male 14.10 68.43 32.68
Other offenses\d
----------------------------------------------------------------------
Female 16.99 63.89 31.28
Male 17.06 64.68 32.15
----------------------------------------------------------------------
\a These probabilities reflect the percent chance that petitioned
cases involved detention of the alleged offender in a secure holding
facility before disposition.
\b These probabilities reflect the percent chance that status-
offender cases petitioned to juvenile courts resulted in formal
adjudication of the juveniles as status offenders.
\c These probabilities reflect the percent chance that cases formally
adjudicated as status-offense cases resulted in the juveniles'
receiving out-of-home placements.
\d The other offenses category includes various status offenses as
defined by individual states, such as tobacco and curfew violations.
Source: Developed by GAO from NCJJ's National Juvenile Court Data
Archive.
Regarding running away, our analyses showed that males had higher
probabilities than females of being detained, adjudicated, or placed.
Further, males with liquor offenses had higher probabilities of being
adjudicated or placed than females. Appendix II provides more
detailed analyses of NCJJ national data.
The national estimates did not enable us to determine whether gender
bias occurred in the outcomes because these data did not contain
variables that are likely to be relevant to judges' decisions (e.g.,
prior offense histories). Accordingly, we developed statistical
models to measure gender bias using data sets that contained
appropriate variables.
--------------------
\11 NCJJ's national data did not include cases not petitioned to the
juvenile court.
ANALYSES OF INDIVIDUAL
STATES' DATA DID NOT
INDICATE BIAS
---------------------------------------------------------- Letter :4.2
To analyze gender bias, we developed logistic regression models of
the intake decisions for six of the seven states and the detention,
adjudication, and placement decisions for seven states. Overall, the
25 models involved applications of the logistic regression procedure.
That is, each state's models contained variables that measured
characteristics that may be associated with the juvenile judicial
system outcomes\12 and estimated how the characteristics influenced
outcomes. These characteristics included the source of referral to
the juvenile court, location (e.g., metropolitan or rural area) of
the court, age and race of the offender, type of offense, and
offender's prior offense history. We used these models to test for
gender bias. For the intake decisions, we analyzed all cases
referring to the intake staff; for the detention and adjudication
decisions, we analyzed only petitioned cases; for the placement
decisions, we analyzed cases of adjudicated status offenders.
Table 2 shows the "gender-bias quotients," which were the resulting
estimates of gender bias from the models that we developed. As the
gender-bias quotient approaches 1.0, the amount of estimated gender
bias decreases. No specific criteria exist as to the extent that the
quotient would have to deviate from 1.0 to indicate gender bias. In
our judgment, however, a deviation from 1.0 of more than .2 would
indicate the presence of gender bias.\13 Our results indicated that
(1) in 5 of the 6 intake models, females were about equally as likely
as males to be petitioned to juvenile court and (2) in 14 of the
other 19 models, no gender bias was demonstrated in the juvenile
justice systems' outcomes for status offenders. In the Florida
intake model, females were more likely to be petitioned to juvenile
court than males because the juvenile justice system treated females'
characteristics, e.g., type of offense, differently due to their
gender. In the other five models, we found some indication of gender
bias in Arizona's, Florida's, and Nebraska's detention decisions and
Florida's adjudication and placement decisions. These models
indicated that females were less likely to be detained or placed than
males because the juvenile justice system treated females'
characteristics, e.g., referrals by the police, differently due to
their gender. Our conclusions about gender bias are limited to
aspects of the juvenile justice process for which we had data. See
appendix III for a detailed explanation of the models and the
methodology.
Table 2
Gender-Bias Quotient Results Regarding
Intake Decisions, Detentions,
Adjudications, and Placements in Seven
States
Adjudicati
State Intake Detention on Placement\a
------------------------ ------ ----------- ---------- -----------
Arizona\b 1.15 3.39 0.99 Not
available\c
California\d 0.94 1.02 0.93 1.07
Florida 0.78 1.34 1.22 2.97
Missouri 0.96\e 0.98 1.02 0.85
Nebraska Not 1.44 1.02 0.80
availa
ble\f
South Carolina 1.06 1.03 1.01 0.80
Utah 0.98 Not 1.00 0.83
available\g
----------------------------------------------------------------------
\a Our placement model covers only those cases formally adjudicated
as status-offender cases.
\b The NCJJ data for Arizona cover only Maricopa County, which
represents about 57 percent of the juvenile population between the
ages of 10 and 17.
\c We were unable to develop a placement model for Arizona because
the state's data file had an insufficient number of placed
status-offender cases.
\d The California data cover counties (Alameda, Los Angeles, San
Francisco, San Joaquin, and Ventura) that represent about 40 percent
of the state's juvenile population between the ages of 10 and 17.
\e Missouri data did not indicate which cases were handled formally.
\f Nebraska intake data were not available for two of its largest
counties.
\g We were unable to develop a detention model for Utah because the
state's data file did not contain detention information.
Source: GAO modeling using NCJJ state-specific juvenile court
records from calendar years 1990 through 1991.
In measuring gender bias, we combined the effects of the individual
variables to estimate the overall probabilities of intake decisions,
detention, adjudication, and placement. By combining these effects
to estimate gender bias, some variables may have had offsetting
effects, regardless of whether the models showed gender bias. For
example, in the Missouri intake results, which did not indicate
gender bias, law enforcement and school referrals for females lowered
their probability of being petitioned, but urban courts increased the
probability of being petitioned. These offsetting situations
occurred relatively infrequently.
Our analysis showed that certain factors, such as offenders' prior
offense history and source of referral, affected the status
offenders' outcomes. For example, as would be expected, offenders'
prior offense history generally affected their detention outcomes.
As the number of prior offenses increased, so did the probability
that the status offenders would be detained regardless of whether
they were females or males. See appendix III for a discussion of the
influence of such characteristics on the intake, detention,
adjudication, and placement outcomes.
--------------------
\12 Data files from the states of Nebraska, Arizona, and Utah did not
contain sufficient information to develop certain models.
Specifically, we were unable to develop an intake model for Nebraska
because two of its largest counties did not report nonpetitioned
cases. Also, we could not develop a placement model for Arizona
because there were too few status-offense cases involving placements
to estimate the model. In addition, Utah's data file did not contain
detention information; therefore, we could not estimate a detention
model.
\13 The use of .2 (20 percent) was derived from criminological
research on racial discrimination. (See app. III for further
discussion.)
SURVEY RESPONSES OF
PROBATION OFFICERS REVEALED
NO DIFFERENCES IN PROCESSING
---------------------------------------------------------- Letter :4.3
Table 3 shows that the probation officers who responded to our survey
did not perceive any differences in the way females and males with
similar status-offense histories were processed. More specifically,
of the responding probation officers, we estimated that
-- 71.6 percent did not report any differences in the
referral/arrest process,
-- 79.1 percent did not report any differences in the intake
process, and
-- 70.5 percent did not report any differences in either treatment
by the court or the length and type of disposition.
Regarding the detention process, 50.1 percent of the chief probation
officers did not report any gender differences. However, another
41.8 percent of the chief probation officers reported "no basis" for
answering this part of the question, thought the question not
applicable, and/or did not answer the question.\14
Table 3
Probation Officers' Responses to Survey
Question Comparing the Processing of
Female and Male Status Offenders
Other
Question 12: In your jurisdiction, responses
what are the differences, if any, (e.g., no
in the ways females and males with basis, not
similar status offense histories No Some applicable,
are processed with respect to each differe difference or no
of the following: nce s response)
---------------------------------- ------- ---------- -------------
Referral/Arrest 71.6 9.2 19.2
Detention\a 50.1 8.0 41.8
Intake 79.1 2.6\b 18.3
Treatment by the court 70.5 9.2 20.3
Length and type of disposition 70.5 6.6\b 22.9
----------------------------------------------------------------------
Note: See appendix V, question 12, for a categorization of the
responses.
\a Total does not add to 100 percent because of rounding.
\b Because of the small number of responses, the sampling errors were
calculated differently (see app. I).
Source: GAO survey.
--------------------
\14 All estimates to the population are subject to sampling errors.
All estimates in this report are within 5 percentage points on either
side of the estimate, with a confidence interval of 95 percent,
unless otherwise noted. For more information, see appendix I.
COMPARATIVE AVAILABILITY OF
FACILITIES AND SERVICES FOR
STATUS OFFENDERS BELIEVED
SIMILAR
------------------------------------------------------------ Letter :5
Generally, both our national survey respondents and the juvenile
justice officials and facility representatives we interviewed in four
states told us there were not any significant differences in the
facilities and services available to female and male status
offenders. However, both groups emphasized that they believed that
more services were needed for status offenders, irrespective of
gender.
GAO NATIONAL SURVEY
RESPONSES INDICATED EQUAL
TREATMENT OPTIONS
---------------------------------------------------------- Letter :5.1
As table 4 shows, 44.4 percent of the chief probation officers who
responded to our survey said that treatment options (facilities and
services) were about equally available to detained female and male
status offenders. However, more than one-third of the respondents--
37.8 percent--reported "no basis" for answering this question,
thought the question inapplicable, or did not answer the question.
Therefore, about 70 percent of those officials who responded said
that the services and facilities were about equal for detained female
and male status offenders.
When the respondents reported a difference, the difference was
generally related to the perception that there were more facilities
available for males than females. Almost 16 percent of the
respondents reported that facilities and services were either
"somewhat more" or "much more" available for males than for females.
In contrast, only 2 percent of the respondents said that facilities
and services were either "somewhat more" or "much more" available for
females than males.
Table 4
Probation Officers' Responses to Survey
Question Comparing the Availability of
Facilities and Services for Detained
Status Offenders
Question 9a: Based on your experience, how would you
describe the current availability of treatment options Percentage
(facilities and services) for detained status offenders? of responses
-------------------------------------------------------- ------------
Much more available for males 6.9\a
Somewhat more available for males 8.9
About equal for females and males 44.4
Somewhat more available for females 1.4\a
Much more available for females 0.6\a
Other responses (e.g., no basis, not applicable, or no 37.8
response)
Total responses 100.0
----------------------------------------------------------------------
Note: See appendix V, question 9a.
\a Because of the small number of responses, the sampling errors were
calculated differently (see app. I).
Source: GAO survey.
Further, many respondents indicated that the availability of
facilities and services for status offenders perhaps would be more
accurately described as being equally unavailable for females and
males. For example, some respondents said that female and male
status offenders had no treatment programs or facilities due to
limited funding and resources. In addition, other respondents said
that the existing services were inadequate to meet the needs of both
genders. Four other respondents to our survey indicated that, even
within an overall environment of limited resources for both genders,
female status offenders had fewer services than males.
SERVICES FOR FEMALE AND MALE
STATUS OFFENDERS IN SELECTED
FACILITIES WERE SIMILAR
---------------------------------------------------------- Letter :5.2
As table 5 shows, we visited a total of 15 facilities--10
co-educational facilities and 5 serving only females or only males.
Except for some health services not applicable to males (such as
prenatal care), we generally did not find gender-based distinctions
in the availability of counseling, educational, and medical services
for females and males at each of the 10 co-educational facilities we
visited. Officials at the other five facilities said that their
programs were not gender-based and could be provided to either
females or males.
OJJDP officials pointed out that providing similar services for both
females and males may be equitable but may not result in meeting the
specific needs of one gender.
Table 5
List of 15 Facilities Visited (by Type,
Name, Location, and Gender Served)
County (urban
Facility type Facility State or rural) Gender served
-------------- -------------- -------------- -------------- ----------------
Detention Regional Florida Duval (urban) Female and male
facility Juvenile
Detention
Center
Juvenile Kentucky Fayette Female and male
Detention (urban)
Center
Big Sandy Kentucky Johnson Female and male
Regional (rural)
Detention
Center
Juvenile Texas Bexar (urban) Female and male
Detention
Center
Shelter Youth Crisis Florida Duval (urban) Female and male
Center South
Interface Florida Alachua Female and male
Runaway (rural)
Shelter
Coleman House Kentucky Fayette Female and male
(urban)
Second Mile Maryland Prince Georges Female and male
Runaway Home (urban)
Walden Sierra, Maryland St. Mary's Female and male
Inc. (rural)
Letot Center's Texas Dallas (urban) Female and male
Emergency
Shelter
Group home Bardstown Kentucky Jefferson Male
Group Home for (urban)
Boys
Project Kentucky Fayette Female
Respect Group (urban)
Home for Girls
Salvation Army Texas Bexar (urban) Female
Adolescent
Treatment
Center
San Antonio Texas Bexar (urban) Female
Youth
Residential
for Females
Practical and Florida Duval (urban) Female
Nonresidential Cultural
program Education
Center for
Girls
--------------------------------------------------------------------------------
Source: Developed by GAO from information provided by facility
officials.
COUNSELING SERVICES
-------------------------------------------------------- Letter :5.2.1
At the 10 co-educational facilities (4 secure detention centers and 6
shelters), we did not find gender-based distinctions in counseling
services offered for female and male status offenders. Generally,
the four secure detention facilities did not routinely provide
counseling to females or males. Facility officials told us that
youth who requested counseling or who displayed suicidal tendencies
were referred to community health-care providers. The officials
added that their facilities basically were temporary holding centers
for youth awaiting juvenile court processing and were not designed to
provide treatment services. According to these officials, while many
of the resident youth may need counseling and mental health services,
the centers were not the appropriate facilities for providing these
services.
At the six shelters, the resident female and male youth were provided
weekly counseling services (individual, group, or both). Individual
counseling, available at all six shelters, ranged from 2 hours to 6
hours per week. Group counseling, available at five of the shelters,
ranged from 4 hours to 14 hours per week.
The other five facilities (four group homes and one nonresidential
program), which served either females or males, also provided
individual counseling (ranging from 1 hour to 4 hours per week) and
group counseling (1 hour to 5 hours per week).
EDUCATIONAL SERVICES
-------------------------------------------------------- Letter :5.2.2
At the 10 co-educational facilities, we did not find gender-based
distinctions in the availability of educational services for female
and male status offenders. Youth at these 10 facilities attended
public schools or on-site schools, with 1 exception.\15
The four group homes, each serving either only females or only males,
sent youth to local public schools, an on-site alternative school, or
alternative schools operated by the state and the local public school
district. The 15th facility (nonresidential program) was an
alternative school and, therefore, provided education on-site.
--------------------
\15 The Big Sandy Regional Detention Center (Johnson County, KY) did
not have on-site educational services for youth who were not
transported to the local public schools.
MEDICAL SERVICES
-------------------------------------------------------- Letter :5.2.3
According to service providers at all 15 facilities we visited,
females and males received needed medical services, either at the
respective facility or from local community health-care providers.
Generally, we did not find gender-based distinctions in the
availability (from either on-site or community sources) of medical
care for females and males at the 10 co-educational facilities,
except for services, such as gynecological services and prenatal
care, which were not applicable to males. Admission physicals were
the only gender-based difference we noted. At two of the female-only
group homes, health examinations included testing (which could be
refused at one of the homes) for sexually transmitted diseases,
whereas, at similar male-only facilities operated by the same
organizations, such testing was not done unless requested by the
males.
Only 5 of the 10 co-educational facilities, 4 detention facilities
and 1 shelter, had on-site medical personnel. Each of these five
facilities had a doctor on-site at least 1 day per week. Also, each
of the four detention centers had a nurse on-site at least 5 days per
week, and the shelter had a nurse on- site 3 days per week. Some of
the on-site service providers told us that their facilities were
often overcrowded and in need of additional medical staff. At two
detention facilities, for example, officials told us that the on-site
nurse could not fully treat all of the females and males on each
day's sick list. According to the officials, the nurse at this
facility had to select which patients to treat. The other five
co-educational facilities (five shelters) did not have doctors or
nurses on site. Residents of these facilities relied on parents,
guardians, or, if necessary, facility staff to provide access to
community health-care services.
The remaining five facilities--three group homes serving only
females, one group home for only males, and the nonresidential
program for females--generally did not have on-site medical personnel
and, thus, relied on community health-care providers. Some officials
at the shelters and group homes that did not have on-site medical
personnel told us that such resources were needed for medical
services. For instance, one official explained that counselors had
to use their already limited counseling time to dispense medication
and transport youth to doctors' offices.
JUVENILE JUSTICE SYSTEM
PERCEIVED AS NOT MEETING
STATUS OFFENDERS' SERVICE
NEEDS
---------------------------------------------------------- Letter :5.3
Juvenile court judges, detention officials, and service providers in
the nine counties we visited said that more facilities and services
were needed for both female and male status offenders. Some of the
juvenile justice representatives and professional staff said that
early intervention services were needed for first-time offenders to
divert them from further involvement with the juvenile justice
system. For example, some judges said that while not all status
offenders become delinquent offenders, the majority of the juvenile
delinquents appearing in their courts had a previous history of
status offenses.
Most of the juvenile justice officials and service providers we
interviewed told us that status offenders did not need
gender-specific treatment or services, except for gynecological
services and prenatal care for females. In fact, representatives
from the female-only and male-only facilities said that their
programs could be replicated to provide the same counseling and
mental health services to status offenders of the opposite sex.\16
Other officials added that gender did not play a role in determining
a youth's individual treatment needs because each youth had unique
needs. Further, some service providers said that facilities should
serve both females and males because the two genders would have to
communicate and interact on a daily basis, such as they would in
real-life situations.
Some service providers pointed out that advantages exist to having
single-gender facilities because distractions or anxieties could be
created when both genders participate in the same counseling and
treatment programs. For example, a service provider at a female-only
facility told us that many of the females had experienced some form
of abuse by males. Thus, according to this provider, a female-only
program was more conducive to helping the females work through their
feelings and build self-esteem. Appendix IV provides more details
about our visits to the selected facilities in the four case-study
states.
--------------------
\16 Illustrations of this replication in Kentucky are the Project
Respect Group Home for Girls and the Bardstown Group Home for Boys
(see table 5). Both of these single-gender facilities were operated
by the same nonprofit organization and provided similar treatment and
services to the resident females and males.
---------------------------------------------------------- Letter :5.4
We are sending copies of this report to the Attorney General; the
Administrator, Office of Juvenile Justice and Delinquency Prevention;
the Director, Office of Management and Budget; and other interested
parties. Copies will also be made available to others upon request.
Major contributors to this report are listed in appendix VI. If you
have any questions about this report, please contact me on (202)
512-8777.
Laurie E. Ekstrand
Associate Director, Administration
of Justice Issues
OBJECTIVES, SCOPE, AND METHODOLOGY
=========================================================== Appendix I
The 1992 reauthorization (P.L. 102-586) of the Juvenile Justice and
Delinquency Prevention Act of 1974 (P.L. 93-415) mandated that we
study gender-bias issues in state juvenile justice systems.
Specifically, we agreed with the Committees to
-- compare the outcomes of intake decisions and frequency of
detentions, adjudications, and out-of-home placements of female
and male status offenders and
-- compare the availability of facilities and services for female
and male status offenders in selected jurisdictions.
In addressing these objectives, we reviewed relevant literature.
Regarding the first objective, we analyzed the frequency of
detentions, adjudications, and out-of-home placements of petitioned
status offenders by gender at the national level, and we made
comparisons within selected states.\1 Further, we analyzed intake
decisions within selected states. Regarding the second objective, we
visited a total of 15 facilities in 4 states. Finally, we obtained
additional perspectives on these juvenile justice issues by mailing a
survey to a national sample of county juvenile justice probation
department officials.
--------------------
\1 We used data that were housed in and made available by the
National Juvenile Court Data Archive, which is maintained by NCJJ and
supported by a grant from OJJDP. These data were originally
collected by the Maricopa County, AZ, Juvenile Court Center; the
Alameda County, CA, Probation Department; the Los Angeles County, CA,
Probation Department; the San Francisco County, CA, Juvenile
Probation Department; the San Joaquin County, CA, Probation
Department; the County of Ventura, CA, Corrections Services Agency;
the Florida Department of Health and Rehabilitative Services; the
Missouri State Division of Children and Youth Services; the Nebraska
Commission on Law Enforcement and Criminal Justice; the Pennsylvania
Center for Juvenile Justice Training and Research; the South Carolina
Department of Juvenile Justice; and the Utah Juvenile Court. Neither
the original data collectors nor NCJJ bear any responsibility for our
analyses or interpretations of the data.
REVIEW OF RELEVANT LITERATURE
--------------------------------------------------------- Appendix I:1
To develop an understanding of gender-bias issues associated with
state juvenile justice systems, we reviewed relevant literature
identified in bibliographies provided us by the National Center for
Juvenile Justice (NCJJ) and the Department of Justice's Office of
Juvenile Justice and Delinquency Prevention (OJJDP). Our review of
the literature aided us in defining gender bias and in designing
models to conduct our analyses of intake decisions, detentions,
adjudications, and out-of-home placements in selected states.
NATIONAL DATA COMPARING
PETITIONED FEMALE AND MALE
STATUS OFFENDERS' FREQUENCY OF
DETENTIONS, ADJUDICATIONS, AND
PLACEMENTS
--------------------------------------------------------- Appendix I:2
To develop national statistics comparing the frequency that female
and male status offenders were detained, adjudicated, and placed, we
used juvenile court data collected annually by NCJJ. Each year, NCJJ
collects juvenile court case-level data from various states and
jurisdictions and assigns weights to the data, which permits
projecting the data to produce national estimates of cases disposed
by all state juvenile justice systems.\2 OJJDP publishes the weighted
data in its annual report entitled Juvenile Court Statistics.
Using NCJJ's data files (the National Juvenile Court Data Archive),
we developed statistics for a 6-year period from calendar years 1986
to 1991.\3 More specifically, we developed national estimates of the
gender-specific probabilities of detentions, adjudications, and
out-of-home placements for petitioned status offenders by offense
categories for the 6-year period and annually.
Our comparative analyses of NCJJ data have some significant
limitations. For example, the NCJJ data did not represent the
universe of status offenders. Rather, the data included only those
status offenders who were petitioned to or otherwise handled more
formally by the juvenile courts. Thus, the data did not include
status offenders who were picked up, counseled, and/or released by
the police. Nor did the data cover those juveniles who received
informal dispositions from county juvenile department officials
during intake screening. For example, intake officials may counsel
and release the juveniles or divert them to social service agencies.
Another significant limitation of our national-level analyses is that
any differences in the resulting frequency and probability statistics
(comparing female and male status offenders in reference to
detentions, adjudications, and placements) cannot be used to draw
interpretations or conclusions about either the presence or the
absence of gender bias.\4 For the purposes of our review, we defined
gender bias as differences in juvenile justice systems' outcomes
(intake decisions, detentions, adjudications, and placements) of
female and male status offenders who had similar characteristics,
such as age, status offense, and offense history. Thus, because
NCJJ's national data files contained insufficient information on
prior offense histories and other variables relevant to judges'
decisions in the cases, we could not use our national-level analyses
to draw interpretations or conclusions about gender bias.
Despite these limitations, the national-level frequency and
probability statistics provide a useful overview regarding petitioned
status offenders.
--------------------
\2 In 1991, for example, the following 23 states provided juvenile
court case-level data to NCJJ: Alabama, Arizona (Maricopa County
only), Arkansas, California, Connecticut, Florida, Maryland,
Minnesota, Mississippi, Missouri, Montana, Nebraska, New Jersey, New
York, North Dakota, Ohio, Pennsylvania, South Carolina, South Dakota,
Texas, Utah, West Virginia, and Wisconsin. In 1990 and 1991,
California reported data from several of its larger counties
representing 40 percent of the state's youth population at risk. In
addition, some jurisdictions in seven other states reported court
case-level data that were used in generating the national estimates.
In all, data from 1,504 jurisdictions covering 57 percent of the
Nation's youth at risk were used to produce the 1991 national
estimates. NCJJ's estimates of the number and characteristics of
delinquency and petitioned status-offense cases disposed by juvenile
courts were on the basis of the assumption that the characteristics
of cases in counties that did not report juvenile court statistics
were similar to those counties of similar size that did report
statistics to NCJJ. The details of the estimation procedures can be
found in OJJDP's annual report entitled Juvenile Court Statistics.
NCJJ's national estimates were not generated by a probability sample.
However, NCJJ has conducted tests of the validity of the national
estimates by comparing their referral estimates to counts of
referrals (as reported by the Federal Bureau of Investigation's
annual Uniform Crime Reports) made by law-enforcement agencies to
juvenile courts. NCJJ concluded that the data were generally
reliable.
\3 At the time of our review, NCJJ did not have national statistics
available for calendar year 1992 or later years.
\4 National estimates for intake decisions were not available.
ANALYSES OF INTAKE DECISIONS,
DETENTIONS, ADJUDICATIONS, AND
PLACEMENTS IN SELECTED STATES
--------------------------------------------------------- Appendix I:3
While NCJJ's national data files did not contain sufficient
information for directly analyzing gender-bias issues, some of the
Center's state-specific files did have a wider range of variables
(including prior offense histories) to permit such analyses. For
example, in addition to gender and type of status offense, some of
the variables relevant to our analyses were: the age of the youth at
the time of referral to the juvenile justice system, the outcome or
finding of the adjudicatory hearing, and whether the youth had any
previous referrals and/or adjudications. Thus, to conduct more
detailed analyses of intake decisions, detentions, adjudications, and
placements, we selected the following 7 states from the total of 25
states that provide data to NCJJ: Arizona, California, Florida,
Missouri, Nebraska, South Carolina, and Utah. In addition to
geographical coverage, we considered the following factors in
selecting these seven states.
-- The states' juvenile justice systems reflected a diverse range
of processes for handling youthful offenders.
-- The states' data files contained a sufficient number of relevant
variables to permit construction of models to test the
respective state's juvenile justice system for indications of
gender bias in the handling of similarly situated female and
male status offenders.
For each of the seven states selected, we obtained a copy of NCJJ's
computerized data files for calendar years 1990 and 1991, the most
recent years for which consistent data were available.\5
Then, using the 1990 and 1991 data files for all status offenders, we
constructed logistic regression\6 models for intake decisions. We
used a class of models commonly used in criminological research to
analyze these types of outcomes. For petitioned status offenders, we
constructed logistic regression models to test for gender-based
differences (if any) in three other aspects of juvenile justice
processing. These models contained variables to measure offenders'
characteristics.
First, we tested how the characteristics affected the probabilities
associated with female and male status offenders' being detained
before adjudication. Second, we tested how the characteristics
affected the probabilities, by gender, of being formally adjudicated
as a status offender. Third, we tested how the characteristics
affected the probabilities of females' and males' receiving placement
as a final disposition. However, we could not address possible
gender bias elsewhere in the juvenile justice system because data did
not exist. For example, the data did not include youths who were
handled informally--that is, picked up, counseled, and/or released by
the police or by county juvenile department intake officials.
Appendix III presents the results of our regression analyses of
intake decisions, detentions, adjudications, and placements.
To gain an understanding of the juvenile justice systems in the seven
states in our analyses, we interviewed state officials in various
jurisdictions within those states, including judges, prosecutors, and
juvenile justice specialists. Those interviews covered many topics,
including the referral process; the prosecution, adjudication, and
disposition of juveniles; the juvenile justice systems in various
jurisdictions; workload; and state laws as they related to the
processing of juvenile offenders.
--------------------
\5 For trend purposes, additional data files (i.e., for years before
1990) would have been desirable; however, the 1990 and 1991 data
files were the only years that had a sufficient range of variables
common to all seven states to facilitate our planned analyses. For
four states, we also obtained data for 1988 and 1989 to obtain
juveniles' prior criminal history records.
\6 Logistic regression analysis is a widely accepted statistical
methodology used when the dependent variable is qualitative, such as
if a status offender is detained. Regression analysis identifies
relationships between the dependent variable and two or more key
variables, such as the gender and age of the juvenile, offender's
current offense category, and prior offense categories and
dispositions. (See app. III for a list of the dependent and
independent variables.)
OVERVIEW COMPARISONS OF THE
AVAILABILITY OF FACILITIES AND
SERVICES FOR FEMALE AND MALE
STATUS OFFENDERS IN SELECTED
JURISDICTIONS
--------------------------------------------------------- Appendix I:4
To develop comparative information about the availability of
facilities and services for female and male status offenders, we
visited a total of nine counties--generally two counties (a rural
county and an urban county) within each of four states (Florida,
Kentucky, Maryland, and Texas).\7 In judgmentally selecting these
states, our primary consideration was that we wanted to visit
juvenile justice jurisdictions that reflected various approaches for
handling status offenders and/or provided a variety of facilities and
services, including some facilities serving only females, some
serving only males, and some serving both genders. Thus, in
selecting states to visit, we first solicited suggestions from
juvenile justice professionals with national or multijurisdiction
experience. These professionals included, for example, OJJDP
officials, as well as representatives of advocacy groups, such as the
Coalition for Juvenile Justice and the National Girls' Caucus.
Following are more specifics regarding our reasons for selecting each
state.
-- Florida had began a process of privatizing services to status
offenders by contracting with the Florida Network of Youth and
Family Services, Inc., which operated residential shelters and
nonresidential treatment and counseling sites throughout the
state.\8 Also, according to OJJDP officials, Florida had a
female-specific program (the Practical and Cultural Education
Center for Girls) that had received national attention.
-- Kentucky, in 1986, had enacted legislation providing for
informal processing of juveniles involved in less-serious
offenses. These juveniles may enter into diversion agreements,
which impose conditions such as community service, counseling,
curfew, and restitution.
-- Maryland tries to divert status offenders from the juvenile
justice system into nonresidential counseling programs operated
by youth service bureaus, which are private, not-for-profit
organizations under contract with the state's Department of
Juvenile Services. Also, according to OJJDP officials, Maryland
was one of only a handful of states that began planning for
gender-specific services for juvenile offenders before such
planning was required by federal legislation.\9
-- Texas is a populous state with a relatively large number of
juveniles. According to 1990 census data, 3 of the 10 most
populous U.S. cities are in Texas. We visited Dallas and San
Antonio, which we selected on the basis of our available staff.
Generally, in deciding which counties to visit in each of the four
states, a primary criterion we used was the relative volumes of
status offenders referred to and/or detained by the local juvenile
justice systems. We obtained referral and detention information by
reviewing (1) periodic reports that county juvenile justice officials
submit to the respective state's office of the governor and (2) each
of the states' current 3-year plans submitted in conjunction with
applications for formula-grant funding under the Juvenile Justice and
Delinquency Prevention Act. Using these data sources and considering
suggestions of state juvenile justice specialists, we selected one
urban and one rural county to visit in each of the four states,
except in Texas, where we selected two urban counties--Dallas County
and Bexar County. We selected two urban counties in Texas because we
wanted to contrast different approaches for dealing with status
offenders. For example, Dallas County had a separate juvenile
probation facility (the Letot Center) specifically designated for
only status offenders, while Bexar County had no such separately
designated facilities. Also, each county had one of the nation's 10
most populous cities--Dallas and San Antonio.
Generally, in each of the selected counties, we interviewed local
juvenile justice officials (judges, law-enforcement officers,
detention facility officials, and others) to obtain overview
perspectives on the availability of facilities and services for
status offenders. Also, we visited facilities that the state and
local officials identified as having services or being placement
options for status offenders. In total, we visited 15 facilities--4
detention facilities, 6 shelters, 4 group homes, and 1 nonresidential
program.\10
At the facilities, we obtained information about the capacity, or
number of beds available; genders served by offense category; extent
of overcrowding, if applicable; and average lengths of stay. Also,
we toured the facilities to obtain information on available
counseling, educational, and medical services--that is, the services
most relevant to the principal needs of status offenders. In
addition, we interviewed the service providers (the professional
staff responsible for providing counseling, educational, and medical
services) at each of the facilities to obtain views on the treatment
needs of status offenders, including views on the need for
gender-specific services. We did not verify the information facility
officials gave to us, nor did we try to evaluate or compare services
provided. The results of our visits cannot be projected to other
counties and facilities within the respective states, and comparisons
should not be made between states.
--------------------
\7 We visited three counties in Kentucky (see table IV.1 in app.
IV).
\8 The Florida Network of Youth and Family Services, Inc., is a
not-for-profit association of agencies and individuals serving
families and youth (ages 10 through 17), including runaways and those
youth at risk of running away; dropping out of school; or becoming
delinquent, abused, neglected, or abandoned. The Network's services
include specialized counseling, safe temporary shelter, food, and
clothing.
\9 The 1992 amendments to the Juvenile Justice and Delinquency
Prevention Act require states applying for grants to submit plans (to
OJJDP) that contain an analysis of available gender-specific services
for females.
\10 The specific counties and facilities we visited in each of the
four states are detailed in appendix IV (see table IV.1).
SURVEY SENT TO A NATIONAL
SAMPLE OF PROBATION DEPARTMENT
OFFICIALS
--------------------------------------------------------- Appendix I:5
We conducted a mail survey of county probation department officials
nationwide to obtain their views on issues concerning gender bias.
At our request, NCJJ gave us a list of all juvenile probation
departments in the United States.\11 NCJJ identified 1,410 officials
whose titles indicated that they were the main officials in juvenile
probation departments. Titles on the list included "chief probation
officer," "court services director," and "court administrator." We
referred to all such individuals as "chief probation officers." The
list of 1,410 officials was developed by eliminating duplicates in
counties listing more than one individual as the chief probation
officer. NCJJ then selected a random sample of 500 such officials
for our sample. Although we sent our survey to the individual
listed, some questionnaires were actually completed by other
individuals in their offices (see app. V). The survey was designed
to (1) identify differences in relationship to gender in the juvenile
justice system's processing of status offenders and (2) obtain
perspectives on the availability of facilities and services for
status offenders. By using a national sample, we were able to
project the results of our study to a population of 1,249 chief
probation officers.
We designed and pretested the survey in March and April 1994 and
mailed it to the 500 randomly selected officials in May 1994. As
needed, we made some follow-up inquiries by mail and/or telephone to
help ensure an adequate response rate. We determined that 57
questionnaires had been sent to offices that did not handle status
offenders; therefore, we eliminated these offices from our sample and
adjusted the universe, accordingly. Our resulting study population
was 1,249 chief probation officers, and our valid sample consisted of
443 such individuals. We received a total of 349 useable responses
out of the 443 surveys mailed, for a response rate of 79 percent.
All such samples are subject to sampling error. All percentage
estimates noted in this report are within plus or minus 5 percentage
points, using a 95-percent confidence interval, with the following
exceptions. The following calculations either exceed the 5-percent
range or are calculated using a Poisson distribution because of the
small number of responses. All sampling errors reported here use the
95-percent confidence interval.
-- Estimate referred to as "About 70 percent" on pages 3 and 15:
71.4 percent, sampling error is 5.8 percent.
-- Estimate of 2.6 percent in table 3: confidence interval for the
percentage is from 1.5 percent to 4.2 percent.
-- Estimate of 6.6 percent in table 3: confidence interval for the
percentage is from 5.0 percent to 8.6 percent.
-- Estimate of 6.9 percent in table 4: confidence interval for the
percentage is from 5.2 percent to 8.9 percent.
-- Estimate of 1.4 percent in table 4: confidence interval for the
percentage is from .7 percent to 2.8 percent.
-- Estimate of .6 percent in table 4: confidence interval for the
percentage is from .2 percent to 1.8 percent.
-- Estimate of 2 percent on page 16: confidence interval for the
percentage is from 1.1 percent to 3.5 percent.
In addition to the reported sampling errors, the practical
difficulties of conducting any survey may introduce nonsampling
errors. For example, variations in the wording of questions, the
sources of information available to the respondents, or the types of
people who do not respond can lead to somewhat different results. We
included steps in both the data collection and data analysis stages
for the purpose of minimizing such nonsampling errors. For example,
we pretested the survey on members of the target population. All
returned surveys were manually edited, double-keyed, and verified for
accurate data entry, and all computer analyses were checked by a
second independent analyst.
--------------------
\11 According to NCJJ officials, the list they gave us covered more
than 99 percent of all juvenile probation departments in the nation.
NATIONAL DATA COMPARING PETITIONED
FEMALE AND MALE STATUS OFFENDERS
========================================================== Appendix II
According to the National Center for Juvenile Justice (NCJJ) data,\1
500,620 status-offense cases were petitioned to juvenile courts in
the United States during the 6-year period from 1986 to 1991. As
mentioned in appendix I, because NCJJ's national data files contained
insufficient information on prior histories and other variables
relevant to judges' decisions in the cases, our national-level
analyses cannot be used to draw interpretations or conclusions about
either the presence or the absence of gender bias. Of the total
petitioned status-offense cases, 206,756 cases (41.3 percent)
involved females and 293,864 cases (58.7 percent) involved males.
These proportions were fairly consistent across the 6 years. (See
tables II.1 and II.2.)
--------------------
\1 NCJJ data are the source of all data in this appendix. These data
represent "cases" and not "individuals." An individual youth may be
involved in more than one status-offense case during any given time
period; that is, the individual may be a repeat offender.
RUNNING AWAY APPEARS TO BE MORE
A FEMALE-RELATED STATUS-OFFENSE
CATEGORY AND LIQUOR OFFENSE
MORE A MALE-RELATED CATEGORY
-------------------------------------------------------- Appendix II:1
In terms of gender distinctions, two specific offense categories with
noticeable differences in the frequency (number) of female and male
status-offense cases petitioned to juvenile court were running away
and liquor offense. Running away appeared to be a predominantly
female category. For the 6-year period shown in tables II.1 and
II.2, females were involved in 61.9 percent of the total 83,000
petitioned running away cases, and males were involved in the other
38.1 percent. In contrast, liquor offense appeared to be a
predominantly male category. Of the total 156,317 petitioned liquor
offense cases during 1986 through 1991, males were involved in 74.3
percent of the cases, and females were involved in the other 25.7
percent.
Table II.1
Number of Status-Offense Cases
Petitioned to Juvenile Courts by Offense
Category, Calendar Years 1986-1991
Status-offense category 1986 1987 1988 1989 1990 1991 Total
------------------------ ------ ------ ------ ------ ------ ------ ------
Liquor offense 24,124 25,112 25,870 23,882 29,049 28,280 156,31
7
Running away 15,580 14,569 12,873 12,168 12,934 14,876 83,000
Truancy 21,698 21,009 21,299 20,950 24,602 25,986 135,54
4
Ungovernable 16,652 14,534 13,405 11,787 11,491 11,228 79,097
Other offenses 6,314 6,799 7,140 7,871 8,822 9,716 46,662
All offenses 84,368 82,023 80,587 76,658 86,898 90,086 500,62
0
--------------------------------------------------------------------------------
Source: Developed by GAO from NCJJ's National Juvenile Court Data
Archive.
Table II.2
Percentage of Status-Offense Cases
Petitioned to Juvenile Courts by Gender,
Calendar Years 1986-1991
Status offense (category
and gender) 1986 1987 1988 1989 1990 1991 Total
------------------------ ------ ------ ------ ------ ------ ------ ------
Liquor offense
Female 23.4 25.7 24.1 26.0 27.1 27.5 25.7
Male 76.6 74.3 75.9 74.0 72.9 72.5 74.3
Running away
Female 61.8 62.2 62.2 61.9 62.2 61.0 61.9
Male 38.2 37.8 37.8 38.1 37.8 39.0 38.1
Truancy
Female 46.6 44.8 46.2 46.0 45.6 46.4 46.0
Male 53.4 55.2 53.8 54.0 54.4 53.6 54.0
Ungovernable
Female 49.8 51.1 48.7 48.2 45.5 47.2 48.6
Male 50.2 48.9 51.3 51.8 54.5 52.8 51.4
Other offenses
Female 29.0 31.8 32.0 31.8 32.0 29.5 31.0
Male 71.0 68.2 68.0 68.2 68.0 70.5 69.0
All offenses
Female 42.1 42.1 40.8 41.2 40.5 41.2 41.3
Male 57.9 57.9 59.2 58.8 59.5 58.8 58.7
--------------------------------------------------------------------------------
Source: Developed by GAO from NCJJ's National Juvenile Court Data
Archive.
PROBABILITIES THAT PETITIONED
STATUS OFFENSE CASES INVOLVED
DETENTION
-------------------------------------------------------- Appendix II:2
During 1986 through 1991, of the total 500,620 status offense cases
petitioned to juvenile courts, 10.7 percent (53,748 cases) involved
secure detention of the alleged offender before disposition.\2 Of the
total detention cases, 43.4 percent (23,326 cases) involved females
and 56.6 percent (30,422 cases) involved males.
Table II.3 presents the results of our probability analyses regarding
the 53,748 cases involving secure detention during 1986 through 1991.
Generally, the probabilities, or percent chances, for females and
males within each respective offense category were similar. For
example, during the 6-year period shown, a female status offender
petitioned for a liquor offense had a 4.96-percent chance of being
detained, compared with a 6.37-percent chance for a male offender.\3
For most offenses, the probability of being detained decreased for
both males and females between 1986 and 1991. For example, the
probability of female runaways' being detained decreased from about
33 percent in 1986 to about 13 percent in 1991; for males, the
percentage dropped from 38 percent to 23 percent.
Table II.3
Comparative Chance (by Status-Offense
Category and Gender) of Petitioned
Status Offenders' Being Detained Before
Disposition by Juvenile Courts, Calendar
Years 1986-1991
Combin
ed
probab
ility,
Status offense (category 1986-
and gender) 1986 1987 1988 1989 1990 1991 1991
------------------------ ------ ------ ------ ------ ------ ------ ------
Liquor offense
Female 7.95 6.44 4.55 2.35 5.81 3.10 4.96
Male 6.48 7.15 5.26 5.86 7.09 6.32 6.37
Running away
Female 32.61 31.27 23.39 19.25 13.80 12.86 22.54
Male 37.95 36.41 28.13 23.03 18.94 22.86 28.28
Truancy
Female 6.26 4.76 2.59 1.98 2.04 2.55 3.31
Male 6.20 6.00 3.25 2.54 2.72 3.01 3.90
Ungovernable
--------------------------------------------------------------------------------
Female 18.94 18.14 14.26 9.11 8.65 7.84 13.61
Male 19.30 18.46 14.53 11.11 9.20 9.29 14.10
Other offenses
Female 19.42 18.55 18.37 10.90 17.37 18.12 16.99
Male 9.25 14.07 19.44 13.22 22.36 20.89 17.06
--------------------------------------------------------------------------------
Source: Developed by GAO from NCJJ's National Juvenile Court Data
Archive.
--------------------
\2 As defined in the glossary of terms applicable to NCJJ data,
secure detention is the "placement of a youth in a restrictive
facility between referral to court intake and case disposition."
However, NCJJ's national-level data do not indicate the period or
length of detention.
\3 Statistically speaking, the numbers presented in tables II.3,
II.4, and II.5 represent conditional probabilities as percentages.
That is, numbers represent the probability of detention given that a
case was petitioned or handled formally by the juvenile courts.
These probabilities are reported separately by gender and by offense
category. For example, the conditional probability of detention for
females referred for liquor violations is 4.96 percent. In other
words, females who committed a liquor offense and were petitioned to
the juvenile courts during calendar years 1986 through 1991 had a
4.96-percent chance of being detained. The following formula shows
the actual calculation of this probability: 1,993.3/40,219.7 equals
4.96 percent. In this formula, which covers combined data for the
6-year period, 1,993 is the total number of liquor-offense cases
involving females who were detained, and 40,220 is the total number
of liquor-offense cases involving females. This applies to all
subsequent tables in this appendix.
PROBABILITIES THAT PETITIONED
STATUS OFFENDERS WERE FORMALLY
ADJUDICATED
-------------------------------------------------------- Appendix II:3
During 1986 through 1991, of the total 500,620 status-offense cases
petitioned to juvenile courts, 62.0 percent (310,363 cases) were
formally adjudicated as status offenders. In these 310,363 cases,
the adjudicatory hearings resulted in formal findings or
determinations of status-offense conduct.\4 Of the 310,363
adjudicated cases, 40.3 percent (124,923 cases) involved females and
59.7 percent (185,440 cases) involved males.
Table II.4 presents the results of our probability analyses regarding
the 310,363 adjudicated cases during 1986 through 1991. Generally,
the adjudication probabilities for females and males within each
respective offense category were comparatively similar. For example,
during the 6-year period shown, a female status offender petitioned
for a liquor offense had a 57.34-percent chance of being adjudicated,
compared with a 59.69-percent chance for a male offender.
Table II.4
Comparative Chance (by Status-Offense
Category and Gender) of Petitioned
Status Offenders' Being Formally
Adjudicated by Juvenile Courts, Calendar
Years 1986-1991
Combined
Status offense probability
(category and , 1986-
gender) 1986 1987 1988 1989 1990 1991 1991
------------------- ------ ------ ------ ------ ------ ------ -----------
Liquor offense
Female 53.53 58.70 57.27 56.33 55.38 61.83 57.34
Male 60.19 57.88 58.98 57.78 59.50 63.40 59.69
Running away
Female 55.03 53.06 52.11 47.08 43.27 42.68 49.04
Male 59.42 56.09 54.87 51.50 47.32 45.12 52.49
Truancy
Female 69.97 68.21 67.64 66.23 65.43 67.75 67.51
Male 73.45 70.42 68.25 67.79 68.41 68.94 69.51
Ungovernable
Female 68.95 67.16 63.60 63.93 63.07 67.58 65.97
Male 71.53 69.63 68.32 67.74 65.41 66.66 68.43
Other offense
Female 64.33 68.71 60.59 67.69 57.28 65.82 63.89
Male 57.31 63.97 66.49 67.55 64.51 66.62 64.68
--------------------------------------------------------------------------------
Source: Developed by GAO from NCJJ's National Juvenile Court Data
Archive.
--------------------
\4 Juvenile courts conducted an adjudicatory hearing in each of the
total (500,620) petitioned cases. As a result of these hearings,
youth in 310,363 of the cases were formally determined to be (and
were processed as) status offenders. In the other 190,257 cases, the
courts decided not to formally adjudicate the youth as status
offenders.
PROBABILITIES THAT ADJUDICATED
STATUS OFFENDERS RECEIVED AN
OUT-OF-HOME DISPOSITION
-------------------------------------------------------- Appendix II:4
During 1986 through 1991, of the total 310,363 adjudicated
status-offense cases in the United States, 18.3 percent (56,725
cases) resulted in out-of-home placement dispositions for the
offenders.\5 Of these 56,725 cases, 42.4 percent (24,077 cases)
involved females and 57.6 percent (32,648 cases) involved males.
Table II.5 presents the results of our probability analyses regarding
the 56,725 out-of-home disposition cases during 1986 through 1991.
Here again, the probabilities, or percentage chances, for females and
males within each respective offense category were comparatively
similar. For example, during the 6-year period shown, a petitioned
female status offender adjudicated in the running away category had a
31.25-percent chance of receiving an out-of-home disposition,
compared with a 34.68-percent chance for a petitioned male.
Table II.5
Comparative Chance (by Status-Offense
Category and Gender) of Formally
Adjudicated Status Offenders' Receiving
Out-of-Home Dispositions by Juvenile
Courts, Calendar Years 1986-1991
Combined
Status offense probability
(category and , 1986-
gender) 1986 1987 1988 1989 1990 1991 1991
------------------- ------ ------ ------ ------ ------ ------ -----------
Liquor offense
Female 6.17 6.79 7.49 5.76 4.41 4.24 5.68
Male 8.45 8.65 7.87 8.80 8.90 7.96 8.42
Running away
Female 33.49 36.03 29.46 29.60 27.20 29.30 31.25
Male 38.42 42.95 34.18 23.94 29.41 34.85 34.68
Truancy
Female 10.27 10.28 8.17 8.98 9.34 8.17 9.17
Male 10.07 12.60 11.28 9.95 9.77 8.80 10.35
Ungovernable
Female 31.69 34.31 32.21 30.75 31.45 31.58 32.11
Male 31.95 33.79 31.40 33.53 36.74 28.82 32.68
Other offenses
Female 29.92 29.53 35.91 33.91 32.19 26.95 31.28
Male 25.79 30.09 31.91 41.16 36.29 26.57 32.15
--------------------------------------------------------------------------------
Source: Developed by GAO from NCJJ's National Juvenile Court Data
Archive.
--------------------
\5 Our analyses of dispositions focused on out-of-home placements
rather than other disposition alternatives, such as dismissal or
probation.
TECHNICAL DISCUSSION OF
GENDER-BIAS MODELS
========================================================= Appendix III
This appendix describes our research to measure gender bias in the
case processing of status offenders in four juvenile justice system
outcomes. These outcomes were: (1) the intake decision to petition
status offenders to juvenile court versus the decision to handle them
informally; (2) the decision to detain petitioned status offenders
securely prior to an adjudicatory hearing; (3) the outcome of an
adjudicatory hearing; and (4) the decision to place adjudicated
status offenders out-of-home in secure or nonsecure placements. We
analyzed 1990 and 1991 juvenile court data from up to seven states or
counties within selected states\1 for each of the four outcomes. We
measured gender bias in these four outcomes as the discrepancy or gap
between females' actual outcomes and the outcomes that they would
have received had they been treated as males were treated.
More specifically, we used juvenile court case-level data to estimate
gender-specific logistic regression equations of the relationships
between each of the four outcomes and case characteristics. That is,
for female and male status offenders, we estimated separate
regressions for whether (1) a case was petitioned at intake, (2) a
case petitioned at intake was detained, (3) a petitioned case was
adjudicated, and (4) an adjudicated case was placed out-of-home.
We included as independent or explanatory variables in our
regressions three types of case characteristics. These
characteristics were: (1) offense-related characteristics, such as
current offense and prior offense history; (2) justice-system
variables, such as the source of referral to the juvenile court, the
location of the court, and, for the adjudication and placement
outcomes, whether the case was detained during its processing; and
(3) offender characteristics, such as age and race. The variables in
the final models were selected from a broader set of variables using
appropriate statistical techniques. The broader set of variables was
identified from the literature on gender bias, but it was limited to
those variables actually available in a given state's database.
We estimated the separate logistic regressions by gender, to derive
gender-specific estimates of the juvenile justice systems' treatment
of females' and males' characteristics.\2 We took these estimates of
the systems' treatment of males' characteristics and applied them to
females' average characteristics to predict females' outcomes if
their characteristics were treated equal to males'.\3 We defined as
gender bias the gap between these two sets of outcomes--i.e., those
models predicted for females versus those that we estimated would
have occurred had females been treated as males.
In general, we found that females received outcomes that were similar
to the ones they would have received if their average characteristics
had been treated like males' characteristics. In only 6 of the 25
models, across the 4 outcomes in the 7 states that we analyzed, did
we find outcomes that we characterize as evidence of gender bias.
Across states, but within case-processing outcomes, we found some
similarities and some differences in the variables that were
associated with the outcomes. For example, prior offense history
tended to be strongly and positively associated with each of the four
outcomes across the states (that had variables measuring prior
offense history). However, the effects of other characteristics on
particular outcomes were not consistent across states. For example,
whether a case was referred to the courts by law-enforcement agencies
was positively associated with the likelihood of detention in
Arizona, California, and Nebraska; but it had no effect on the
likelihood of detention in Florida, Missouri, and South Carolina.
The models alone do not explain why these outcomes may happen. For
example, the difference between states may be due to differences in
police procedures, police practices, or laws.
We found similarities and differences across the states in the
characteristics of females and males who were processed by their
juvenile justice systems. Across states, males tended to have more
prior contacts with the juvenile justice system than females, and
males also tended to be slightly more likely to be referred to intake
by law-enforcement agencies than females. Also across the states, we
found gender differences in the types of offenses for which status
offenders were referred to juvenile courts. There tended not to be
differences between males and females on the basis of age and race.
Finally, within and across states and outcomes, we found some gender
differences in the courts' treatment of individual characteristics.
Specifically, we found cases in which variables had opposite effects
on the likelihood of an outcome for females than on the likelihood of
that same outcome for males. For example, in California, females
referred to the court by law-enforcement agencies were less likely to
be petitioned to juvenile court than females referred by other
sources; however, males referred to the courts by law-enforcement
agencies were more likely to be petitioned than males referred by
other sources. In general, however, the direction of the effects of
variables were consistent between the females' and males' equations.
That is, the same variables that increased or decreased the
likelihood of a particular outcome for females also tended to
increase or decrease the likelihood of that particular outcome for
males. In addition, we had cases in which a variable influenced an
outcome for one gender, but not the other gender.
--------------------
\1 The states included Arizona, California, Florida, Missouri,
Nebraska, South Carolina, and Utah. The data from Arizona were
limited to cases processed in Maricopa County. The data from
California came from five counties: Alameda, Los Angeles, San
Francisco, San Joaquin, and Ventura. Two of the largest counties in
Nebraska did not report nonpetitioned cases. Therefore, we did not
analyze the intake decisions in Nebraska.
\2 We followed the specification adopted by Samuel L. Myers, Jr., in
"Statistical Tests of Discrimination," Journal of Quantitative
Criminology (Vol. 1, No. 2, pp. 191 to 218, 1985).
\3 The method we used to estimate gender bias is discussed in more
detail later in this appendix.
SCOPE OF THE ANALYSIS
------------------------------------------------------- Appendix III:1
We analyzed calendar year 1990 and 1991 juvenile court case-level
data for up to seven states for each of four case processing
outcomes. The outcomes were (1) whether a case was petitioned by
intake staff, such as juvenile probation officers, to juvenile court
for more formal handling or hearing by a judge; (2) whether a case
petitioned to juvenile court was detained before its formal
hearing;\4 (3) whether petitioned cases were adjudicated as status
offenders; and (4) whether adjudicated cases were placed out-of-home.
Table III.1 reports the number of cases used in the analysis for each
stage. Table III.2 reports the proportion of female cases in each
stage. The number of cases referred in table III.1 represents the
total sample of cases coming into the juvenile justice systems in
each state, that is, cases referred from law-enforcement officers,
schools, family, social service agencies, and other sources. From
the cases referred, a subset is petitioned at intake to juvenile
court (the number petitioned). Of those petitioned, a subset is
detained (the number detained), and a subset is adjudicated as status
offenders (the number adjudicated). Finally, of those cases
adjudicated, a subset is placed out-of-home (the number placed).
The data in table III.1 show that the number of cases referred to the
respective juvenile justice systems ranged from almost 41,000 in
Missouri to about 8,700 in the 5 California counties. The number of
cases processed at each of the other stages--detention, adjudication,
and placement--also varied across the states.
Table III.2 shows the proportion of females at each stage for each
state. These proportions varied by outcomes and states. For
example, in Utah, about 30 percent of the cases referred to the
juvenile courts were females, whereas, in South Carolina, about 49
percent of the cases referred were females. Similar ranges and
variability across the states occurred in other stages of processing.
Table III.1
Number of Status-Offender Cases Referred
to Intake, Petitioned to Juvenile Court,
Detained, Adjudicated, and Placed Out-
of-Home, by State, 1990-1991
Cases Cases Cases Cases Cases
referr petiti detain adjudicat place
State ed oned ed ed d
---------------------------- ------ ------ ------ --------- -----
Arizona\a 11,801 847 67 370 14
California\b 8,757 910 97 570 255
Florida 20,429 3,294 162 875 138
Missouri 40,986 4,822 740 3,843 1,394
Nebraska n/a\c 2,489 156 2,245 393
South Carolina 10,576 4,715 120 4,259 158
Utah 20,399 6,109 n/a\d 4,980 192
----------------------------------------------------------------------
\a Arizona data were for Maricopa County only.
\b California data were for five counties--Alameda, Los Angeles, San
Francisco, San Joaquin, and Ventura.
\c Two of the largest counties in Nebraska did not report the number
of cases handled informally; therefore, we did not report the total
number referred.
\d Utah did not report whether cases were detained.
Source: GAO analysis of NCJJ data.
Table III.2
Proportion of Status-Offender Cases That
Were Female Cases Referred to Intake,
Petitioned to Juvenile Court, Detained,
Adjudicated, and Placed Out-of-Home, by
State, 1990-1991
Cases Cases Cases Cases Cases
referr petiti detain adjudicat place
State ed oned ed ed d
---------------------------- ------ ------ ------ --------- -----
Arizona\a .3381 .5419 .2836 .5378 .2143
California\b .3251 .3593 .1753 .3684 .3098
Florida .4523 .3257 .2778 .3143 .3571
Missouri .4024 .3932 .3797 .3872 .4060
Nebraska n/a\c .4210 .4182 .3737 .4733
South Carolina .4898 .4674 .5000 .4628 .5127
Utah .3040 .2734 n/a\d .2707 .2760
----------------------------------------------------------------------
\a Arizona data were for Maricopa County only.
\b California data were for five counties--Alameda, Los Angeles, San
Francisco, San Joaquin, and Ventura.
\c Two of the largest counties in Nebraska did not report the number
of cases handled informally; therefore, we did not report the total
number referred.
\d Utah did not report whether cases were detained.
Source: GAO analysis of NCJJ data.
In table III.3, we report the gender-specific aggregate probabilities
for each of our four decision points by state. The following
probabilities were defined:
-- the probability of being petitioned at intake equals the number
of cases petitioned to juvenile court divided by the number
referred to the intake office,
-- the probability of secure detention equals the number of
petitioned cases detained securely divided by the number of
petitioned cases,
-- the probability of adjudication equals the number of petitioned
cases adjudicated as status offenders divided by the number of
petitioned cases, and
-- the probability of placement equals the number of adjudicated
cases receiving an out-of-home placement divided by the number
of adjudicated cases.
Table III.3
Gender-Specific Probabilities of
Petition, Detention, Adjudication, and
Out-of-Home Placement, by State, 1990-
1991
Petiti Detent Adjudicati
State Gender on ion on Placement
------------ ------------- ------ -- ------ ---------- ---------
Arizona Female .1150 .0414 .4335 .0151
Male .0497 .1237 .4407 .0643
California Female .1149 .0520 .6422 .3762
Male .0986 .1372 .6175 .4889
Florida Female .1161 .0419 .2563 .0836
Male .1985 .0527 .2701 .1917
Missouri Female .1150 .1482 .7848 .3804
Male .1195 .1569 .8048 .3516
Nebraska Female n/a\a .0544 .8960 .2737
Male n/a\a .0687 .9063 .1792
South Female .4255 .0272 .8943 .0411
Carolina
Male .4653 .0239 .9112 .0336
Utah Female .2693 n/a\a .8072 .0393
Male .3127 n/a\a .8182 .0383
----------------------------------------------------------------------
\a Not available.
Source: GAO analysis of NCJJ data.
As in table III.1, table III.3 shows that there was a wide
variability across states in the probabilities at each stage. There
were also gender differences within states in the probabilities at
particular stages. For example, the probability of being petitioned
at intake to juvenile court for females ranged from about 11 percent
in Arizona, California, Florida, and Missouri to about 42 percent in
South Carolina. Within states, there were gender differences in the
probability of being (1) petitioned at intake, in Arizona and Utah;
(2) detained, in Arizona and California; (3) adjudicated, in
Nebraska; and (4) placed, in Arizona and Nebraska.
Alone, differences in these aggregate probabilities did not reveal
gender bias. The probabilities did not account for gender-specific
differences in the distribution of case characteristics that were
associated with each of the outcomes. The presence or absence of
gender differences in the probabilities may mask gender differences
in case characteristics or gender differences in the manner in which
the respective juvenile justice systems treated the characteristics.
Gender differences in the treatment of characteristics could lead to
gender bias in outcomes. For example, the absence of a large gender
difference in the probability that cases were petitioned at intake to
the court in Missouri (.1195 for males as compared to .1150 for
females) could mask gender bias or gender differences in treatment.
If, for example, intake offices in Missouri were more likely to
petition male liquor offenders than female liquor offenders, but
female liquor-law violators comprised a larger portion of the sample
of female cases, then the aggregate probabilities of being petitioned
at intake may mask the difference in treatment on similar
characteristics.
--------------------
\4 In general, the data did not contain measures on the length of
time cases were detained.
TECHNICAL ASPECTS OF ESTIMATING
AND MEASURING GENDER BIAS IN
JUVENILE COURT OUTCOMES
------------------------------------------------------- Appendix III:2
We measured gender bias as the gap or discrepancy between females'
outcomes as determined by their average characteristics and females'
outcomes under the assumption that their average characteristics were
treated the same as males. We devised a measure--the gender-bias
quotient--to summarize the degree to which these two sets of outcomes
differed.
The gender-bias quotients were developed from the results of the
gender-specific regressions of each of the four outcomes. In
general, we estimated separate models for females and males using the
case characteristics as predictors or independent variables.\5 Upon
estimating the regressions, we produced parameter estimates of the
influence on a dependent variable of each of the independent
variables. For each outcome, we had two sets of parameter estimates,
one for females and one for males.
We used the parameter estimates and the case characteristics for
females and males to construct the gender-bias quotients. To do so,
we calculated two sets of predicted average probabilities for
females. The first predicted probability we called the "model
probabilities." These were the predicted average probabilities for
females for each outcome, e.g., the probability of being petitioned
to juvenile court. The model probabilities were calculated using the
mean or average characteristics of females in the sample. To compute
the model probabilities, we multiplied the female parameter estimates
for each independent variable by the respective means of the
independent variables for females. We summed across these products
and transformed the result into a probability to produce the model
probabilities.\6
The second probability we calculated was the "equal treatment"
probability. We followed a similar procedure as above. However, in
this case, we multiplied the parameter estimates for males by the
average characteristics of females, summed the products, and
transformed the result into the "equal treatment" probabilities. The
ratio of the equal treatment to the model probabilities yielded the
gender-bias quotient.
The gender-bias quotient measures the extent to which females'
outcomes diverge from males' outcomes if case characteristics were
treated equally. The gender-bias quotient is an aggregate measure in
that it is produced by summing across the effects of different
variables. It is possible, therefore, that the aggregate gender-bias
quotients may show little or no gender bias, but that there may be
gender differences in treatment on particular variables. The results
of our regression analysis enabled us to identify situations where
there were differences in treatment on particular variables but no
aggregate gender bias, as measured by the gender-bias quotients.
Further, the method we used to construct the gender-bias quotients
takes into account two sets of influences on each of the four
case-processing outcomes. The first influence is the differences in
the average characteristics of female and male status offenders
across all cases. The second influence is the differences in how
females' and males' characteristics were treated. Discrepancies
between the two sets of predicted probabilities that comprise the
gender-bias quotients arising from the first set of influences are
not indicators of gender bias; those discrepancies arising from the
second set are indicators.
The distinction between these influences stems from the fact that the
outcomes we reviewed--petitioned at intake, detention, adjudication,
and placement--may be determined by a number of variables, such as
current offense, prior offense history, age, and race. If some
variables had larger influences on these outcomes than others and the
variables with larger influences were correlated with gender, then
there would be gender differences in these outcomes. Such
differences would not be characterized as gender bias, however,
because they are explained by the gender differences in the
distribution of case characteristics. Failure to control for gender
differences in case characteristics may lead to the incorrect
inference that there is gender bias in the outcomes, when, in fact,
what has been observed is gender differences in the distribution of
variables associated with outcomes.
On the other hand, estimated differences in the way the juvenile
justice system evaluates females' and males' characteristics, apart
from the distribution of these characteristics across cases, would
indicate gender bias. That is, differences in the magnitude or
direction of the influence of variables between females and males,
regardless of the distribution of these variables between females and
males, indicate that there is gender bias.
For example, suppose, regardless of gender, that the probability of
being detained before adjudication increases with the number of prior
contacts with the juvenile justice system. Everything else being
equal, if a larger proportion of the sample of males had prior
contacts, or if males had more prior contacts on the average than did
females, then one would expect the probability of detention to be
higher for males than females. This type of result would not
indicate gender bias.
However, if males had as many prior contacts with the juvenile
justice system as females, but males with prior contacts were more
likely than females with prior contacts to be detained, all else
being equal, then gender differences in the probability of detention
arising from this situation would indicate gender bias.
The methodology we employed enabled us to distinguish between these
two sources of influences on the outcomes we analyzed. We were able
to (1) evaluate the extent to which the distribution of
characteristics differed between females and males and (2) measure
whether there were gender differences in the juvenile justice
systems' treatment of these characteristics.
--------------------
\5 The details about the regression specifications and the measures
of the independent variables are discussed below.
\6 The details of this procedure are discussed in the next section.
MODELING STRATEGY FOR
ESTIMATING GENDER BIAS
------------------------------------------------------- Appendix III:3
To assess gender bias we estimated separate regressions for females
and males for each of the four decision outcomes in the seven states.
We fit the regressions on a state-by-state basis using variables that
measured case characteristics in each state's data set. We imposed
as few restrictions as possible on our representations of each
state's juvenile justice system; in other words, each state's
regressions may have had a different number of variables.
The four dependent variables in our analysis--whether a case was
petitioned to juvenile court, detained, adjudicated as a status
offense, or placed out-of-home--were dichotomous. Our ultimate
interest was in the gender-specific probabilities of status
offenders' being petitioned at intake, detained, adjudicated, and
placed. This posed two problems. First, the dichotomous dependent
variables violated the assumptions underlying the classical, or
linear, regression model. Specifically, the errors were
heteroskedastic. Second, we wanted to use the regression results to
predict aggregate, gender-specific probabilities for our outcomes,
rather than simply predict the outcomes in individual cases.
The problems posed by the nature of the dependent variables and the
need to estimate probabilities were solved by using a logistic
specification for the regressions. This specification is commonly
chosen by criminologists who analyze data containing dichotomous
outcomes, such as whether a case was convicted.
Using a logistic specification to estimate the parameters, we took
the following three steps to estimate parameters and calculate
predicted probabilities. First, by state, we estimated the separate
regressions for each of the four dependent variables. We included
specific variables in the regressions by assessing the adequacy of
the models both in terms of the individual variables and from the
point of view of the overall fit of the model to the data. In
general, we sought to build the most parsimonious models consistent
with the data, but we also attempted to include theoretically
relevant variables--such as the type of status offense--where
possible.\7
Second, we used the regressions to generate the first set of
predicted probabilities for females and males for each of the four
outcomes. We labeled these probabilities "p(1)," that is, "model
probabilities." These probabilities were calculated from the
gender-specific parameter estimates of the influence of case
characteristics on outcomes and the gender-specific distribution of
case characteristics. Specifically, we multiplied the means\8
of females' characteristics by the parameter estimates from the
female equations, and we summed across the products to estimate the
logit of the probability of interest. These were transformed into
estimated probabilities by the formula:
In this formula, "p" is the predicted probability of an outcome,
e.g., the probability of petitioned at intake; "e" refers to the
operation of exponentiating; and "logit(p)" is the estimated logit of
the probability of the particular outcome. The logit was evaluated
at the mean levels of the variables in the regression equation.
These probabilities, the model probabilities, indicated how females
and males, respectively, were treated by the courts on their average
characteristics.
Third, we used the parameter estimates from the males' equations to
estimate outcomes for females if they were treated in the same way as
males. These probabilities were labeled "p(2)" or the "equal
treatment probabilities" for females. We computed these
probabilities by multiplying the means of the females' variables by
the parameter estimates from the males' equations. We used these
products to predict the equal treatment probabilities for females.
Finally, we took the ratio of the two sets of probabilities--"equal
treatment" to "predicted," or p(2) to p(1)--to estimate the
gender-bias quotient. As the gender-bias quotients approach 1, the
amount of gender bias diminishes. Gender-bias quotients greater than
1 indicate that females were less likely to receive a particular
outcome than if their characteristics were treated as males'
characteristics. Gender-bias quotients less than 1 indicates the
reverse, that females were more likely to receive an outcome than if
their characteristics were treated equally to males' characteristics.
For example, a hypothetical outcome of .7 detentions in a state would
suggest that females were more likely to be detained in that state
than males with similar characteristics; an outcome of 1.3, on the
other hand, would indicate that females were less likely to be
detained than males with similar characteristics.
--------------------
\7 For an approach to model building and assessing goodness of fit,
see Applied Logistic Regression, David W. Hosmer and Stanley
Lemeshow (New York: John Wiley & Sons, 1989).
\8 Using the means of the independent variables was one option for
estimating probabilities. Other options included selecting points
along the distribution of a variable.
MODEL SPECIFICATION
----------------------------------------------------- Appendix III:3.1
The general form of our logistic regressions was as follows. If we
denote any one of the dichotomous dependent variables, for example,
detention by D, then the probability of detention, conditioned on a
vector of case characteristics X and a vector of effects B, is given
by
The case characteristics included in the models included variables
that measured offense history, current offense, etc., as described
before. The entire set of variables used in building the models is
reviewed below. Because we estimated separate models for females and
males, the parameters indicate the gender-specific treatments of each
gender's characteristics by the courts.
To estimate the separate logistic regressions, we used maximum
likelihood techniques and obtained the estimated effects for females
and males. From these, and the mean values of the independent
variables, we calculated the estimated probabilities. Continuing
with the example, we calculated p(1) as the probability of detention
for females. We then calculated a second probability of detention,
the equal treatment probability, or p(2):
Here the Bs represent the estimated parameters from the males'
equations, and the Xs are the means of the independent variables for
females.
Finally, the ratio of p(2) to p(1) yielded our measure of
gender-bias. A ratio of 1 indicates no gender bias. The ratio of 1
also is equivalent to a test of no net differences in the entire set
of coefficients between the females' and males' equations. A
gender-bias quotient greater than 1 indicates females are less likely
to have the outcome of interest than are similarly situated males;
conversely a gender-bias quotient less than 1 indicates that females
are more likely to have the outcome of interest than similarly
situated males.
There is no absolute standard for determining how much of a deviation
from 1 in the gender-bias quotients constitutes gender bias. We
determined that deviations greater than plus or minus .2 provided
indications of gender bias. This figure was derived from
criminological research on racial discrimination in which conclusions
that discrimination was not widespread were on the basis of
unexplained differences of up to 20 percent in the outcomes between
black and whites.
The parameter estimates of the effects of independent variables
represent the change in the log of the odds of a dependent variable.
For the purposes of analyzing the magnitude of effects of independent
variables, the change in the log of the odds (or log odds) as a
result of a unit change in an independent variable is difficult to
interpret. However, by exponentiating the parameter estimates, odds
ratios can be calculated. The odds ratios can be interpreted in a
relatively straightforward manner. The odds ratio is an estimate of
how much more likely, or unlikely, it is for the outcome of interest
to be present among those having a particular characteristic than
those not having that characteristic. For example, an odds ratio of
4 for a variable indicating whether a status offender had prior
dispositions would be interpreted to indicate that status offenders
with prior dispositions are four times as likely as those without
prior dispositions to have the outcome (e.g., detention) of
interest.\9
Finally, the method we used to estimate gender bias was based on
methods developed by economists to measure discrimination in labor
markets. Their method, called the "residual difference," measures
discrimination, or bias, in terms of the differences between the two
sets of outcomes after the effects of all relevant variables have
been accounted for. In the residual difference method, a bias or
discrimination is the residual that cannot be explained by the
variables in the model. The strength of the method lies in its
ability to account for bias in terms of the differences in treatments
on characteristics. The major weakness of the method lies in using
an incomplete or incorrect set of variables to estimate the
regressions. Depending on how they are correlated with the outcome
variables, omitted variables or incorrectly included variables could
reduce or increase the "residual difference." Thus, misspecified
models could lead to incorrect inferences about bias.
--------------------
\9 We used the odds ratios to conduct our analysis of gender
differences in the effects of variables. These results are discussed
in the final section of this appendix.
VARIABLES USED IN THE MODELS
----------------------------------------------------- Appendix III:3.2
We fit state-specific models for each decision, using the relevant
variables available in the states' data sets. We used our knowledge
of each state's system to supplement our model-building. In general,
we used five common categories or classes of independent variables to
build our models. These categories included (1) variables to measure
the current offense, prior offense history, and juvenile justice
system contact, such as source of referral for the current offense,
detention prior to adjudication, and personal attributes, such as age
and race, and (2) variables to measure the location and geographic
characteristics of the court.
We defined our four dependent variables as follows:
-- Intake decision: A dichotomous variable to indicate whether a
status offense case referred to a juvenile court's intake office
was petitioned to the court for formal processing.
-- Detention: A dichotomous variable to indicate whether a
petitioned status offender was detained securely before
adjudication.
-- Adjudication: A dichotomous variable to indicate the outcome of
an adjudicatory hearing, specifically, that a case was
adjudicated as a status offender.
-- Placement: A dichotomous variable to indicate whether an
adjudicated status offender was given an out-of-home placement.
The specific variables that fell within the categories of our
independent variables were as follows:
-- Current offense: We used a set of indicator (dichotomous or
dummy) variables to indicate whether the current offense, i.e.,
the referral offense, was for running away, truancy,
ungovernability, liquor-law violations, or other status
offenses.
-- Prior offense history: We used a number of measures of prior
offense history, including the number of prior juvenile court
referrals for any offense over the life of the juvenile, the
number of prior status-offense dispositions during the 2 years
before the current referral, the number of prior delinquency
offense dispositions during the 2 years prior to the current
referral, and the number of prior delinquency adjudications over
the life of the juvenile. Not all measures were available for
each state.
-- Source of referral: We used a set of dummy variables to
indicate the source of referral. The variables for sources of
referral included the law-enforcement agency, school, family,
and other sources. We varied the reference category by state.
-- Age at referral: We used the age of the status offender at the
time that the case was referred.\10
-- Race of the offender: We used two dummy variables to indicate
whether a status offender (1) was black or (2) belonged to
another race or ethnic group.
-- Metropolitan status of the court of venue: Except in
California,\11 we measured the metropolitan status of the court
by a dummy variable to indicate whether a court was located in a
county belonging to a metropolitan statistical area or a primary
metropolitan statistical area. We also measured the population
density per square mile of the county containing the court.
-- Detention status: For the adjudication and placement decisions,
we used a dummy variable to indicate whether a case was detained
securely.
--------------------
\10 The California data did not include age at referral; however, the
data did include age at disposition, which we used as a proxy for age
at referral.
\11 All of the counties in California were metropolitan. To measure
variability across counties in California, we used dummy variables to
indicate cases disposed in each county.
MODEL FITTING PROCEDURES
----------------------------------------------------- Appendix III:3.3
We fit the models to the data on a state-by-state and
outcome-by-outcome basis. We developed models containing a set of
independent variables that fit the data better than other
combinations of independent variables in a state's data set. Across
states, our models did not necessarily contain the same subset of
variables. As a result, we were not able to directly compare the
size of the effects of different variables across states; although we
did attempt to identify which variables in each state's models had
the biggest effects and to make general comparisons about the effects
of variables.
SAMPLE SELECTION CRITERIA
FOR EACH OUTCOME
----------------------------------------------------- Appendix III:3.4
For the petitioned at the intake decision, we used the sample of all
status-offender cases referred to the intake office in a state. We
did not estimate a model of the intake decision for Nebraska because
data on cases handled informally were not reported for the state's
two largest counties. For the detention and adjudication outcomes,
we used the sample of all cases handled formally or petitioned to the
juvenile court. We did not estimate a detention model for Utah
because its data set did not contain measures of detention. For the
detention and adjudication outcomes, we also measured the current
status offense as the referral offense. In estimating the placement
outcomes, we restricted our analysis to those status-offense cases
adjudicated as status offenses. For the placement outcomes, we
measured the offense as the disposed offense. We did not estimate
placement models for status offenders in Arizona because there were
too few cases.
ANALYSIS OF GENDER BIAS
------------------------------------------------------- Appendix III:4
Our findings on gender bias are summarized on table III.4. A
discussion of our results pertaining to the analysis of the
differences in the effects of individual parameters and of offsetting
effects follows the discussion of the gender-bias quotients.
Table III.4 shows, by gender for each state, three results for each
of the dependent variables: (1) the "models probability" of having
been petitioned at intake, detained securely before adjudication,
adjudicated as a status offender, and placed out-of-home, or p(1),
for females and males; (2) the "equal treatment probability" of the
same outcomes for females, or p(2); and (3) the gender-bias quotient,
or ratio of probabilities for females if treated like males, to
females as predicted by the model--i.e., p(2) to p(1).
Table III.4
Summary of Estimated Probabilities and
Gender-Bias Quotients
Femal Femal Femal Femal
State Male e Male e Male e Male e
----------- ----------- ----- ----- ----- ----- ----- ----- ----- -----
Arizona Model 0.018 0.039 0.018 0.002 0.435 0.431 NA\d NA\d
probabilit 0 2 9 3 1 5
y\a
Equal 0.045 0.007 0.428 \ \NA\d
treatment 2 7 7
probabilit
y\b
Gender- 1.15 3.39 0.99 NA\d
bias
quotient\c
California Model 0.074 0.076 0.061 0.015 0.671 0.684 0.500 0.327
probabilit 0 5 8 8 6 8 7 4
y\a
Equal 0.071 0.016 0.639 0.350
treatment 9 2 7 4
probabilit
y\b
Gender- 0.94 1.02 0.93 1.07
bias
quotient\c
Florida Model 0.134 0.083 0.046 0.031 0.262 0.200 0.139 0.034
probabilit 8 7 4 0 3 8 5 8
y\a
Equal 0.065 0.041 0.244 0.103
treatment 2 5 5 3
probabilit
y\b
Gender- 0.78 1.34 1.22 2.97
bias
quotient\c
Missouri Model 0.100 0.101 0.111 0.107 0.828 0.814 0.322 0.360
probabilit 8 8 8 4 7 4 5 6
y\a
Equal 0.097 0.105 0.829 0.307
treatment 7 2 8 6
probabilit
y\b
Gender- 0.96 0.98 1.02 0.85
bias
quotient\c
Nebraska Model NA\d NA\d 0.052 0.035 0.922 0.911 0.070 0.090
probabilit 9 1 6 3 7 3
y\a
Equal \ NA\d 0.050 0.925 0.071
treatment 7 7 9
probabilit
y\b
Gender- NA\d 1.44 1.02 0.80
bias
quotient\c
South Model 0.407 0.357 0.009 0.009 0.938 0.924 0.009 0.012
Carolina probabilit 0 7 9 4 1 8 3 6
y\a
Equal 0.378 0.009 0.932 0.010
treatment 6 7 4 1
probabilit
y\b
Gender- 1.06 1.03 1.01 0.80
bias
quotient\c
Utah Model 0.278 0.239 NA\c NA\c 0.831 0.826 0.020 0.015
probabilit 2 9 6 3 7 5
y\a
Equal 0.236 \ NA\c 0.828 0.012
treatment 3 9 9
probabilit
y\b
Gender- 0.98 NA\c 1.00 0.83
bias
quotient\c
--------------------------------------------------------------------------------
\a The model probability is the aggregate probability for females and
males as predicted by the gender-specific regressions when females'
and males' characteristics were evaluated at their mean levels.
\b As defined in the text, the equal treatment probability, for
females, is the probability of an outcome if females' characteristics
were treated equal to males' characteristics. As in the case of the
model probability, females' characteristics were evaluated at their
mean levels.
\c The gender-bias quotient is the ratio of females if treated like
males' probability to the females' predicted probability.
\d Data were not available.
Source: GAO analysis of NCJJ state data.
OVERVIEW OF FINDINGS ON
GENDER BIAS
----------------------------------------------------- Appendix III:4.1
In analyzing the gender-bias quotients, we were interested in whether
the aggregate outcomes for females differed from what they would have
been if their characteristics were treated equally to males. If
there were differences, as indicated by gender-bias quotients that
deviated from 1, then we wanted to determine which variables in the
models explained the differences, as previously discussed. Of
secondary concern were those cases in which the gender-bias quotients
were not different from 1, but there were differences in the
treatment of specific characteristics between females and males.
In most of the outcomes we analyzed across the seven states, there
was little evidence of widespread gender bias. In other words, for
most of the outcomes, the gender-bias quotients were near 1. This
was the case in five of the six petitioned-at-intake decision models,
four of the six detention models, six of the seven adjudication
models, and five of the six placement models. In other words, across
a diverse set of states, which represented different types of
juvenile justice systems, females and males tended to receive similar
treatment.
The exceptions to this general finding occurred in the following
decision points: (1) in petitioning-at-intake decisions, females in
Florida were estimated to be more likely to be petitioned to juvenile
court than if they were treated equal to males; (2) in detention
decisions, females in Arizona,\12 Florida, and Nebraska were
estimated to be less likely to be detained than males; (3) in the
adjudication decision, females in Florida were estimated to be less
likely to be adjudicated than males in that state; and (4) in the
placement decisions, females in Florida were estimated to be less
likely to be placed than males in Florida.
In addition, while only Florida's placement outcome deviated by more
than .2 from a gender-bias quotient of 1, in two other states,
Nebraska and South Carolina, the gender-bias quotients for the
placement decisions were .80. In addition, in two other states,
Missouri and Utah, the gender bias quotients were less than 1 and
near .8. Overall, in four of the six states where placement data
were available, the gender-bias quotients for the placement decisions
were less than 1. While only the result for Florida was consistent
with our definition of gender bias, in these other four states, there
appeared to be a slightly higher likelihood for placing females
out-of-home as compared to similarity situated males, but the
magnitude of the effect in any of these four states was not large
enough to lead us to conclude that there was significant gender bias.
The odds ratios from the parameters of the regression models provided
some insight into the reasons for gender bias in the cases identified
above. In the petitioning decisions in Florida in which females were
more likely to be petitioned to juvenile court than in their
equal-treatment outcomes, the gender differences in treatment arose
around female runaways and in the location of the juvenile courts.
Female runaways were more likely to be petitioned to court than male
runaways; however, female runaways were less likely to be petitioned
than female truants or liquor-law violators. In addition, female
runaways comprised a larger portion of female cases than male
runaways did of male cases. Females in metropolitan areas were about
a third more likely to be petitioned than their male counterparts.
Thus, the higher aggregate likelihood of females to be petitioned to
juvenile court appeared to be due largely to differences in treatment
of female runaways, who also happened to comprise a larger share of
all female status offenders.
In the detention decisions in which females were less likely to be
detained than if they were treated like males, the gender differences
appeared to arise from two different sources: the source of referral
and the type of status offense (in the Arizona case), a variety of
variables (in the Florida case), and the type of status offenders
petitioned to the court (in the Nebraska case). In Arizona,
petitioned females who were referred to the court by law-enforcement
officers were one-tenth as likely to be detained than their male
counterparts. In addition, male status offenders referred by
law-enforcement officers comprised a larger proportion of the sample
of all male status-offender cases than occurred among all female
status-offender cases. Finally, female runaways were more likely to
be detained than male runaways.
In Florida's detention outcomes, gender differences in treatment of
characteristics occurred in a number of variables. Female runaways
and liquor-law violators were less likely to be detained than males
referred to juvenile court for these offenses, and females processed
in metropolitan areas also were less likely to be detained than
males.
In Nebraska's detention outcomes, the gender bias arose because of
gender differences in the treatment of particular types of status
offenders. In particular, females picked up for truancy, liquor, and
other offenses were estimated to be less than half as likely to be
detained than male truants. On the basis of their other
characteristics, females and males were treated about equally.
In Florida's adjudication decision in which females were less likely
to be adjudicated than their equal-treatment outcomes, the type of
status offense was related to the gender bias. Specifically, female
runaways were about three times less likely to be adjudicated than
were male runaways, and females petitioned for liquor offenses were
about one-fifth as likely to be adjudicated as males petitioned for
liquor offenses.
Finally, in Florida's placement outcome, which had the gender-bias
quotient that deviated the farthest from 1, and in which females were
less likely to be placed than their equal-treatment outcomes, the
type of status offense also seemed to be associated with the gender
bias. Specifically, females adjudicated for liquor offenses,
truancy, or ungovernability all were less likely to be placed than
comparable males with these offenses. In addition, females
adjudicated for liquor violations, truancy, and ungovernability were
less likely to be placed than females adjudicated for running away.
Finally, females' prior offense histories were not treated as
severely as males, that is, females with prior offenses were not as
likely to be placed as males with prior offenses.
The lower likelihood of placement for females in Florida does not
necessarily mean that females were better off or that males were
treated more harshly than females. To determine this, it would be
necessary to determine the range of treatment options associated with
various placements. For example, a concern expressed in our site
visits related to the treatment options available or unavailable when
status offenders were placed out-of-home. Placements may be used for
a variety of purposes, including providing services and protecting
females from becoming victims of abuse. This latter concern may be
reflected by the fact that in Florida female runaways were more
likely to be placed than other types of female status offenders.
--------------------
\12 In Arizona, while the gender-bias quotient was relatively large,
3.39, the estimated model probabilities and the estimated equal
treatment probabilities were very small in magnitude. The model
probability was .0023, and the equal treatment probability was .0077.
In other words, the difference in treatment between females and males
was comparatively small.
EFFECTS OF SPECIFIC VARIABLES
ON THE PETITIONED AT INTAKE,
DETENTION, ADJUDICATION, AND
PLACEMENT OUTCOMES
------------------------------------------------------- Appendix III:5
In addition to using the results of the regressions to explain the
occurrences of gender bias, we analyzed the regressions to identify
the variables that were associated with each of the outcomes.
Although differences in the way variables were measured and in the
way states processed status offenders prevented us from making direct
comparisons between the states on each set of models, we did assess
the magnitude of the effects of the variables to identify
similarities and differences.
VARIABLES AFFECTING THE
PETITIONED AT INTAKE
DECISION
----------------------------------------------------- Appendix III:5.1
Across the six states where intake data were available, no single
variable had consistent effects on the decision to petition status
offenders at intake, although prior contact with juvenile court
generally increased the likelihood that a case would be detained. In
four of the six states, the type of status offense for which females
and males were referred to the courts did have a strong association
with the likelihood that the cases were petitioned to the juvenile
courts. Specifically, in California, Florida, and Utah, liquor-law
violators and truants were estimated to be more likely, whether they
were female or male, to be petitioned to the courts than other types
of status offenders. In Arizona, this was true only for truants;
moreover, black males were more likely to be petitioned to court than
black females. In California and Missouri, blacks of either gender
were more likely to be petitioned than persons of other races.
Finally, in Arizona, California, Missouri, and South Carolina, the
source of referral influenced the likelihood that a case was
estimated to be petitioned at intake. In particular, in South
Carolina, cases referred to intake by family members were estimated
to be more likely to be petitioned for both females and males than
were cases referred to intake by other sources.
VARIABLES AFFECTING
DETENTION OUTCOMES
----------------------------------------------------- Appendix III:5.2
No variables had consistent effects across all seven states.
However, when the measures of prior offense history--whether measured
as prior referrals, adjudications, or delinquencies--were available
in a states' data set, the prior offense history tended to be
positively associated with the likelihood of detention for both
females and males. The only exception occurred in the effect of
prior status-offense dispositions on the Arizona detention
probabilities. For males in this case, the number of prior
status-offense dispositions during the 2 years before the current
offense decreased the probability of detention.
Other variables that had large positive effects on the probability of
detention included the source of referral and the particular types of
status offenses. Specifically, cases referred by law-enforcement
agencies were estimated to be more likely to be detained for both
females and males in Arizona, California, and Nebraska. In Arizona
and California, the gender more likely to be detained given that a
case was referred by law- enforcement officials differed. In
Arizona, males referred by the police were about 14 times more likely
to be detained than females referred by the police. In California,
females referred by the police were about 9 times more likely to be
detained than males referred by the police. In South Carolina,
females referred to the court by family members or by schools were
estimated to be more likely to be detained than males referred by
those sources, and females referred by family members and schools
were more likely to be detained than females referred by other
sources.
Female runaways were estimated to be more likely to be detained than
other types of status offenders in Arizona, Florida, and Nebraska.
On the other hand, in Florida, male runaways were more likely to be
detained than were female runaways.
Demographic variables, such as age and race, did not exhibit
consistent effects on detention outcomes across states. However, in
three states, race was associated with the likelihood of detention,
and the effects of race varied with gender. Specifically, in
Arizona, black females were more likely to be detained than black
males; conversely, in Florida and California, black males were more
likely to be detained than black females. In Nebraska,
blacks--female or male--were more likely than whites to be detained.
VARIABLES AFFECTING
ADJUDICATION OUTCOMES
----------------------------------------------------- Appendix III:5.3
Adjudication outcomes for females and males tended to be affected
most by three variables: detention, source of referral, and type of
status offense. In general, detention before adjudication lowered
the estimated probability of adjudication. The estimated direction
of the effects of law-enforcement agencies as a source of referral
tended to change between the detention and adjudication decisions.
Law-enforcement referrals were estimated as more likely to be
detained but less likely to result in cases' being adjudicated as
status offenders. Further, this change in the direction of effects
between detention and adjudication also occurred for status offenders
who were referred for running away. Runaways, in general, were
estimated as less likely to be adjudicated than liquor-law violators;
this was despite the fact that runaways were estimated to be more
likely to be detained than liquor-law violators.
These opposing effects between the two stages of the process may
indicate that the juvenile courts use detention and adjudication in
different ways. It is possible that detention may be viewed as
analogous to a disposition for status offenders. The court may view
detention as a sufficient treatment, given that a youth was warned or
counseled about his behavior, and the court may not view additional
sanctions as necessary. The effects of running away may also be
explained in this manner. Runaways may be more likely to be detained
to give officials time to contact the family and return the juvenile.
These cases then may be less likely to be adjudicated because the
juveniles would have been returned to their families.
Other variables included in the models did not exhibit similar
general trends across the states. For example, a prior offense
history increased the probability of adjudication in three states,
and metropolitan status decreased the probability of adjudication in
three states. The effects of age and race were not consistent for
females and males. These other variables may not have had a
statistically significant effect on the adjudication outcome, or they
may have had a statistically significant effect for females or males
but not both, or the direction of the effects may have varied across
states. In addition, the size of the effects of these variables was
small, raising doubts about their overall impact on adjudication
outcomes. The variations in the patterns for these variables attest
to the differences in the states' processes.
VARIABLES AFFECTING
PLACEMENT OUTCOMES
----------------------------------------------------- Appendix III:5.4
With the exception of a prior offense history and the type of
offense, the relationships between the independent variables and the
placement outcomes were similarly difficult to characterize between
the female and male equations and across the six states' models. A
prior offense history for example, was positively associated with the
likelihood of placement for both females and males in four of the six
states where placement data were available, while in a fifth state, a
prior offense history was positively related to placement for males
but not statistically significant for females.
Other variables, such as the source of referral and type of status
offense, were associated with the likelihood of placement, but the
particular source of referral and type of offense that affected
placement varied across states. For example, in Missouri, cases
referred to the court by the law-enforcement agencies and schools
were less likely to be placed regardless of their gender than cases
referred by other sources. However, in Nebraska, males referred by
family members were less likely to be placed than males referred by
other sources, while females referred by family members were more
likely to be placed than females referred by other sources.
Alternatively, in South Carolina, males whose cases were referred by
schools and family members were more likely to be placed, while
females whose cases were referred by these sources were less likely
to be placed than females referred by other sources.
The placement outcomes for runaways were similar to those of the
adjudication of runaways. In the states in which the type of status
offense was associated with placement, runaways were less likely to
be placed than other types of status offenses. Otherwise, the status
offenses more or less likely to be associated with placement outcomes
varied across the states.
STATE-SPECIFIC RESULTS
----------------------------------------------------- Appendix III:5.5
Finally, when the case processing outcomes within states were
analyzed, there were a few variables that had consistent effects
across outcomes within states. For example, in South Carolina, the
source of referral influenced each outcome--particularly when the
source was school or family. Cases referred by family members,
regardless of gender, were more likely to be petitioned to court than
cases referred by other sources. At the detention stage, the effects
of source of referral varied with gender. Males referred by schools
and family were less likely than females referred by these sources to
be detained. Conversely, at the placement stage, males referred by
schools and family were more likely to be placed than females
referred by these sources.
Second, in some states, the type of status offense influenced the
outcomes, but the effects differed. For example, in California,
truants--regardless of gender--were more likely to be petitioned at
intake than liquor-law violators or runaways. However, at the
adjudication and placement stages, male runaways were less likely to
be adjudicated or placed than female runaways; but both female and
male truants were about equally less likely to be adjudicated or
placed than other types of status offenders.
In general, the effects of specific variables differed across states
and stages of processing. These differences may be due to
differences among the states in the structure or objectives of
juvenile courts.
OVERVIEW OF SERVICES FOR FEMALE
AND MALE STATUS OFFENDERS IN
SELECTED FACILITIES
========================================================== Appendix IV
Generally, we found no significant gender-based differences in the
counseling, educational, and medical services provided to females and
males at the 15 facilities we visited, although the extent of such
services varied by type of facility. However, a majority of the
juvenile justice officials and all of the service providers in the
counties we visited said that more facilities and services were
needed for status offenders, both females and males.
BACKGROUND INFORMATION ON
FACILITIES VISITED
-------------------------------------------------------- Appendix IV:1
Table IV.1 presents background data about each of the 15 facilities
we visited--4 secure detention centers, 6 shelters, 4 group homes,
and 1 nonresidential program. We did not make any comparisons or
evaluations regarding the quality of services for these facilities.
Table IV.1
GAO Visits to Selected Facilities in
Four States (by Urban and Rural
Counties) With Services for Status
Offenders
State and county (urban or
rural) Facility Facility type Age Gender Offender type Capacity Purpose of facility
---------------------------- ------------ -------------- ---------- -------------- --------------- -------------- ----------------------------
Florida
-----------------------------------------------------------------------------------------------------------------------------------------------------
Duval County Practical Nonresidential 14-17 Female Status 60 students To provide an alternative to
(urban) and Cultural , nonsecure offenders and incarceration or
Education delinquents institutionalization for
Center for troubled girls by offering
Girls them academics, independent
life skills training,
counseling, and goal
setting. (The Center also
accepted dependent,
pregnant, or parenting
girls.)
Regional Detention 12-18 Female Status 113 beds To provide for the safety,
Juvenile facility, and offenders care, and custody of
Detention residential, male and juveniles from the time they
Center secure delinquents are detained until their
cases are processed through
the juvenile court.
(According to facility
officials, under Florida
law, status offenders who
are the subject of a
judicial order requiring
detention can be placed in a
juvenile detention center.)
Youth Crisis Shelter, 10-17 Female Status 20 beds To provide shelter and
Center South residential, and offenders counseling to runaway and
nonsecure male homeless youth.
Alachua Interface Shelter, 10-17 Female Status 16 beds To provide temporary shelter
County Runaway residential, and offenders and counseling to runaways
(rural) Shelter nonsecure male to help them and their
families resolve their
conflicts.
Kentucky
-----------------------------------------------------------------------------------------------------------------------------------------------------
Fayette Coleman Shelter, 0-17 Female and Status 16 beds To provide a temporary, out-
County House residential, male offenders and of-home placement
(urban) nonsecure delinquents alternative for children
when secure detention is not
appropriate. (The Coleman
House also provided services
to dependent, abused, and
neglected children.)
Juvenile Detention 7-17 Female and Status 25 beds To provide for the care and
Detention facility, male offenders and custody of youth pending
Center residential, delinquents their release by the
secure juvenile court. (According
to facility officials, under
Kentucky law, status
offenders who violate a
court order can be placed in
a secure juvenile detention
or holding facility.)
Project Group home, 13-17 Female Status 8 beds To provide court-ordered
Respect residential, offenders and residential placement, which
Group Home nonsecure delinquents includes counseling and
for Girls\a educational services, and
promote a positive change in
the girls' negative
behaviors. (Ungovernable
behavior was the most common
status offense referral.)
Jefferson Bardstown Group home, 15-17 Male Status 8 beds To provide court-ordered
County Group Home residential, offenders and residential placement, which
(urban) for Boys\a nonsecure delinquents includes counseling and
educational services, and
promote a positive change in
the boys' negative
behaviors. (Truancy was the
most common status offense
referral.)
Johnson Big Sandy Detention 12-17 Female Status 10 beds To provide for the care and
County Regional facility, and offenders and custody of youths pending
(rural) Detention residential, male delinquents their release by the
Center secure juvenile court. (According
to facility officials, under
Kentucky law, status
offenders who violate a
court order can be placed in
a secure juvenile detention
or holding facility.)
Maryland
-----------------------------------------------------------------------------------------------------------------------------------------------------
Prince Second Mile Shelter, 12-17 Female Status 7 beds To provide short-term
Georges Runaway Home residential, and male offenders and residential shelter,
County nonsecure delinquents including assessment,
(urban) counseling, and educational
services, for runaway,
homeless, or abused youth.
St. Mary's Walden Shelter, 12-17 Female Status 6 beds To provide a protective,
County Sierra, Inc. residential, and offenders temporary living arrangement
(rural) nonsecure male and counseling to runaways
to help them resolve the
problems in their homes.
Texas
-----------------------------------------------------------------------------------------------------------------------------------------------------
Bexar Juvenile Detention 10-17 Female Status 96 beds To provide a secure,
County Detention facility, and offenders and temporary facility for
(urban) Center residential, male delinquents juveniles waiting to appear
secure in court or until placements
can be arranged.
Salvation Group home, 13-17 Female Status 16 beds To provide specialized
Army residential, offenders and clinical services, including
Adolescent nonsecure delinquents individual and group
Treatment counseling, for females.
Program (The Program also provided
services to dependent
females.)
San Antonio Group home, 13-17 Female Status 8 beds To provide substance abuse
Youth residential, offenders treatment and rehabilitation
Residential nonsecure services to medically
for indigent youth.
Females\b
Dallas Letot Shelter, 10-16 Female and Status 24 beds To divert status offenders
County Center's residential, male offenders from juvenile detention,
(urban) Emergency nonsecure reunite them with their
Shelter families whenever possible,
and prevent them from
committing more serious
offenses and progressing
further into the juvenile
justice system.
-----------------------------------------------------------------------------------------------------------------------------------------------------
\a The Volunteers of America of Kentucky, Inc., operates both the
Bardstown Group Home for Boys in Jefferson County, KY, and the
Project Respect Group Home for Girls in Fayette County, KY.
\b The Mexican American Unity Council, Inc., operates the San Antonio
Youth Residential for Females in Bexar County, TX. The Council also
operates a similar male-only group home, the Youth Male Residential
Treatment Facility, that has 14 beds in Bexar County.
Source: Developed by GAO from information provided by facility
officials.
The four secure detention centers held females and males for short
terms in physically restrictive environments pending juvenile court
action. Staff at the four detention facilities told us that the
majority of the youth held were males. In addition, the detention
officials reported that most females and males detained at the
detention facilities were delinquent offenders, not status
offenders.\1 Staff at three of the four detention facilities reported
having problems with overcrowding caused by too many referrals of
female and male youth.\2 For status offenders held over 24 hours, the
detention facilities' staff reported that the average length of stay
ranged from 7 days to 30 days.
Also in the four secure detention facilities, female and male status
offenders could be placed in the same living areas with the more
serious offenders. These serious offenders included delinquents who
may have committed homicide, sexual assault, robbery, or aggravated
assault.\3 Staff told us the youth placed in the facilities were
separated primarily by gender because the detention facilities were
generally overcrowded or had limited bed space. After gender, one
detention facility considered the youths' physical sizes and ages in
making placement decisions within the female-only and the male-only
living areas. For example, the younger, smaller males were not
placed in the same living area with the older, larger males. Staff
at another detention center told us, however, that they had no
flexibility beyond gender in placing females because the facility had
only one living area for females, whereas there were six living areas
for males.\4 Staff said that since most of the referrals received at
the facility were delinquent males, only one living area was set
aside for females.
Of the 11 nonsecure facilities, the 6 shelters provided short-term
care to females and males. Staff at the six shelters told us the
majority of youth served were status offenders. At five of the six
shelters, staff reported serving more females than males. Staff at
the remaining shelter reported serving, on average, an equal number
of female and male youth. Staff at two shelters also said that the
shelters sometimes experienced overcrowding caused by too many female
and male referrals, especially during the months that the local
schools were in session. According to staff at the six shelters, the
reported average lengths of stay for female and male status offenders
ranged from 4 days to 45 days.
Gender was the primary factor in determining living arrangements at
the six co-educational shelters. Female and male status offenders
were not commingled with serious juvenile offenders because the
shelters served only status offenders, less serious delinquent
offenders, and dependents.\5
Of the other five nonsecure facilities, the four group homes provided
long-term care with access to community resources and programs.
Three of the four group homes served only females, and one served
only males. Staff at the two group homes in Texas told us the
majority of the females served were status offenders and/or
dependents. The staff at the male-only group home and the
female-only group home in Kentucky said the facilities served more
delinquent offenders than status offenders. The staff at the four
group homes also said their facilities were not overcrowded because
youth were not accepted unless a bed was available. According to
these staff, the average length of stay for female and male status
offenders ranged from 182 days (about 6 months) to 274 days (about 9
months).
The one nonsecure, nonresidential program for status offenders that
we visited was the Practical and Cultural Education Center for Girls,
located in Jacksonville, FL.\6 The Center's program, which has been
nationally recognized,\7 was not overcrowded because a female student
was accepted only if classroom space was available. A waiting list
was maintained to place females as space became available. According
to officials of this program, the average length of attendance for
females in the program was 243 days (about 8 months).
--------------------
\1 The secure facilities we visited in Duval County (Florida) and in
Fayette and Johnson Counties (Kentucky) detained only those status
offenders who had been ordered into detention by the juvenile courts
for violating a valid court order. The Bexar County Detention Center
(Texas) detained all female and male youth referred to the facility.
Bexar County detention facility officials reported, however, that few
status offenders were detained over 24 hours. At the time of our
March 1994 visits, 7 percent of the female and male youth at the
Duval Detention Center (Florida) were status offenders, and 40
percent of the female and male youth at the Big Sandy Regional
Detention Center (Kentucky) were status offenders. None of the youth
at the Fayette County Detention Center (Kentucky) was a status
offender. Bexar County officials were unable to estimate the number
of status offenders at the facility.
\2 An official at the Big Sandy Regional Detention Center (Kentucky)
said the Center did not have a problem with overcrowding because the
facility's 10 beds were enough to meet the demands of the rural area.
\3 To preclude such commingling of status and delinquent offenders,
in 1979 Dallas County, TX, established an alternative placement
facility, the Letot Center, which housed only status offenders (both
females and males). Before 1979, status offenders in Dallas County
were taken by law-enforcement officers to the local juvenile
detention center, which primarily served delinquent offenders. At
the time of our review, law-enforcement officers took all status
offenders to the Letot Center where their cases were processed and
family members were contacted. The Letot Center counselors, the
status offenders, and the offenders' family members all have input in
determining if shelter care and/or nonresidential counseling is
needed. (Tables IV.1 and IV.2 provide additional background
information on the Letot Center and its services.)
\4 Each of the 7 living areas, or modules, had 12 single-occupancy
rooms.
\5 Dependents are youth who have been placed under the care and
custody of the state because they have been abused, neglected, or
abandoned by their parents or guardians.
\6 The Center at Jacksonville could serve 60 students. Additional
Center nonresidential programs are located in four other cities in
Florida--Bradenton, Fort Lauderdale, Miami, and Orlando.
\7 For example, OJJDP has recognized the successes of the programs.
See "P.A.C.E. Center for Girls: The Florida Program Which for the
Past Eight Years Has Successfully Given Troubled Females a Second
Chance," Profile, 1991, (Vol. 5, No. 5). Profile is published by
Community Research Associates (Champagne, IL), which is under
contract with OJJDP.
GENDER-SPECIFIC INFORMATION
ABOUT COUNSELING, EDUCATIONAL,
AND MEDICAL SERVICES AT
FACILITIES VISITED
-------------------------------------------------------- Appendix IV:2
At each of the 15 facilities visited, we obtained gender-specific
information about counseling, educational, and medical services, that
is, the services most relevant to the principal needs of status
offenders. The results of our visits are summarized below and in
table IV.2.
COUNSELING SERVICES
------------------------------------------------------ Appendix IV:2.1
Female and male status offenders did not routinely receive individual
or group counseling at the four secure detention facilities. These
facilities, however, could obtain counseling services from community
resources if staff or resident youth (including female and male
status offenders) requested such services.\8 Juvenile court judges
could also order the facilities to provide counseling. For example,
professional staff at Florida's Duval County Juvenile Detention
Center told us that juvenile court judges sometimes ordered the
detention center to undertake social assessments and provide
counseling services to female and male status offenders placed in the
facility. The detention center officials said that status offenders
were transported to community health-care providers to receive these
services.
All six shelters, the four group homes, and the nonresidential
program provided a variety of on-site counseling services to
individuals, groups, or both. Female and male status offenders,
however, were provided the same types and amounts of counseling
within the co-educational facilities in which they were placed,
according to officials at the facilities. Counseling topics could
cover physical and sexual abuse, as well as substance abuse issues.
Individual counseling ranged from 2 hours to 6 hours per week at the
shelters, 1 hour to 4 hours per week at the group homes, and 5 hours
per week at the nonresidential program. Group counseling ranged from
4 hours to 14 hours per week at the shelters,\9 3 hours to 5 hours
per week at the group homes, and 1 hour per week at the
nonresidential program. All of these facilities had arrangements
with community health-care providers to supply additional counseling
when needed.
Some facility staff told us that female and male status offenders
needed family counseling, but such service was difficult to maintain
or provide. For example, two of the shelters offer family
counseling, but the programs reportedly were poorly attended. Staff
from two of the group homes said that they could not offer family
counseling because court-ordered placements resulted in youth coming
from all areas of the state. These officials explained that family
counseling was impractical because the parents would have been unable
to attend the sessions since they did not live close to the
respective facility.
Nineteen of the 34 juvenile justice officials and 6 of the 15 service
providers we interviewed emphasized that family counseling is
essential because female and male status offenders were running from
some form of abuse or neglect at home. According to these officials,
family counseling could help correct poor parenting skills, which is
a contributor to abuse and neglect. Staff at one of the group homes
we visited told us that limited resources were used most effectively
only when the whole family was included in the treatment plan.
According to the staff, a group-home facility could build a youth's
self-esteem and correct negative behaviors, but frequently the youth
may be released from the group home and returned to the environment
that caused the negative behaviors. The staff said that in these
situations, where the family issues had not been addressed, the youth
was likely to revert to negative behaviors.
Staff at other facilities told us that parents and guardians did not
always give female and male status offenders the support needed to
address and solve problems. For example, an official at one shelter
said they were unable to return a pregnant runaway to her home
because her single-parent mother was using drugs and had just been
evicted from their apartment.
--------------------
\8 One secure detention facility had a counselor on staff. The
counselor did not provide individual or group counseling to the youth
on a regular or scheduled basis but did respond to requests for
counseling services.
\9 One of the six shelters and one of the four group homes did not
provide formal group counseling.
EDUCATIONAL SERVICES
------------------------------------------------------ Appendix IV:2.2
The 15 facilities we visited provided a variety of educational
services. At three of the four secure detention facilities, status
offenders generally attend on-site schools staffed by licensed
teachers.\10 The other secure facility, Kentucky's Big Sandy Regional
Detention Center, did not have an on-site school. A representative
from the detention center told us resident youth are provided
educational services when the juvenile court judges order the public
schools to transport the youth to their classes.\11
At the six (co-educational) shelters, we found no differences in the
educational services provided to female and male status offenders.
Status offenders at four of the six shelters either attended the
local public schools or received daily or part-time instruction at
the respective facility. These youth generally did not attend the
local schools if they had dropped out of school, were studying for
their general equivalency diploma, or did not reside in the county
where the shelter was located. At the fifth shelter, all female and
male status offenders attended an on-site school staffed by licensed
teachers.\12 At the sixth shelter, all female and male status
offenders attended local schools.
Female and male status offenders also received similar educational
services at the four gender-specific group homes. For example, we
visited one male-only group home and one female-only group home in
Kentucky that were operated by the same organization. Both of these
group homes sent the youth to county alternative schools.\13 The two
remaining group homes (each serving females only) sent resident youth
to the local public schools.\14
Education was a main component of the services offered status
offenders at the nonresidential program we visited in Florida.
Licensed teachers provided basic instruction, which enabled the youth
to earn high-school credits that would aid them in returning to the
public schools or obtaining a general equivalency diploma. Classes
were conducted on the campus of the local community college, which
gave the youth access to other educational services as well.
--------------------
\10 In addition to an on-site school, the Fayette County Juvenile
Detention Center (Kentucky) had a school-release program that allowed
some female and male youth to attend local schools during the day. A
juvenile court judge must approve each youth's participation in the
school-release program.
\11 At the time of our visit in March 1994, none of the three females
and seven males detained at the holding center was attending school.
The official we interviewed told us 4 of the 10 youth (1 female and 3
males) were status offenders.
\12 The on-site instruction provided at the five shelters varied from
4 hours to 5 days per week. One of these five shelters hired a
retired teacher to provide on-site instruction. Youth at the other
four shelters were taught by licensed teachers.
\13 The alternative schools were for students who could not attend
the local public schools because they needed treatment services
and/or had exhibited delinquent behaviors.
\14 One of the three female-only group homes also had an on-site
school staffed by a licensed teacher. The on-site school served
females who had discipline problems, had dropped out of the local
public schools, or were preparing to obtain a general equivalency
diploma.
MEDICAL SERVICES
------------------------------------------------------ Appendix IV:2.3
According to service providers at the 15 facilities we visited,
females and males were receiving needed medical services, either
provided through arrangements by parents or guardians or from local
community health-care providers. For example, pregnant females
admitted to some facilities received prenatal care. Facility staff
at one shelter told us that a male had been referred to a local
dermatologist for severe acne. In addition, staff at several
facilities reported that many of the females and males had to be
referred to community dentists because the youth had never received
dental care before arriving at the facilities.
Also, females and males reportedly were given health screenings
and/or physical examinations before or after admission. The health
screenings included a list of questions to determine each youth's
immediate health needs. The physical examinations typically involved
a nurse' taking each youth's temperature and blood pressure and
checking for any signs of physical distress.
All four of the secure detention facilities and one of the six
shelters had on-site medical personnel. These personnel ranged from
a nurse, who was available from 3 days to 7 days per week, to a
doctor, who was available from 1 day to 5 days per week. Although
these five facilities had on-site medical personnel to address minor
medical problems or dispense prescription medication, some service
providers at these facilities told us that their facilities were
often overcrowded and needed additional medical staff. For example,
at two secure detention facilities, the on-site nurse could not fully
treat all of the youth on each day's sick list and, thus, had to
select patients.
The remaining five shelters, four group homes, and the nonresidential
facility did not have on-site medical services. Four of the 10
officials at these facilities told us that such resources were
needed. For instance, one official explained that counselors were
having to use their already limited counseling time to dispense
medication and transport youth to doctors' offices.
Table IV.2
Summary Observations of Services for
Female and Male Status Offenders at the
15 Facilities Visited
Facility Name/ Gender
type Location served Counseling Educational Medical
----------- ----------- ----------- ----------- ----------- ---------------
Secure Regional Female and Regular on- Youth On-site medical
detention Juvenile male site attended an services
Detention individual on-site included a
Center, and group school, nurse,
Duval counseling which had available 5
County, FL were not six days a week,
provided. classrooms and a doctor,
Youth who and eight available 2
displayed licensed days a week.
suicidal teachers Youth were
tendencies provided by given a
or the county. physical within
requested Daily co- 3 days after
counseling educational admittance to
services classes the facility.
were followed a The most common
referred to basic health problems
community curriculum were colds and
health- that sore throats.
care included
providers. language
arts,
mathematics
, science,
and social
studies.
The center
also
provided
drug
education
and general
equivalency
diploma
preparation
.
Secure Juvenile Female and Regular on- Youth On-site medical
detention Detention male site attended services
Center, individual the on- included a
Fayette and group site nurse,
County, KY counseling school, available 5
were not which had days a week,
provided. one and a doctor,
Youth who classroom available 5
displayed and one days a week.
suicidal licensed Youth were
tendencies teacher given a
or provided by physical within
requested the county. 3 days after
counseling Daily co- admittance to
services educational the facility.
were classes The most common
referred to followed a health problems
community basic were colds and
health- curriculum. sexually
care The center transmitted
providers. also diseases.
provided
remedial
education.
Classes
could be
canceled
when the
center was
overcrowded
.
Secure Big Sandy Female and Regular on- The center On-site medical
detention Regional male site did not services
Detention individual have a included a
Center, and group school. nurse,
Johnson counseling Some youth available 5
County, KY were not attended days a week,
provided. the local and a doctor,
Youth who public available 1 day
displayed school when a week. Youth
suicidal the were given a
tendencies juvenile physical 7 days
or court after
requested judges admittance to
counseling ordered the the facility.
services schools to The most common
were transport health problems
referred to the youth were headaches.
community to their
health- classes.
care
providers.
Secure Juvenile Female and Individual Youth On-site medical
detention Detention male and group attended services
Center, counseling the on- included a
Bexar were not site nurse,
County, TX regularly school, available 7
provided which had days a week,
unless the six and a doctor,
youth classrooms available 3
displayed and six days a week.
suicidal teachers Youth were
tendencies provided by given a
or the county. physical 7 days
requested Daily after
the classes admittance to
services. followed a the facility.
The basic The most common
facility curriculum. health problems
had one The center involved
counselor also sexually
on staff to provided transmitted
meet these independent diseases.
requests. -living
skills
instruction
and
remedial
education.
Classes
could be
canceled
when the
center was
overcrowded
.
Shelter Youth Female and Youth The shelter The shelter did
Crisis male received did not not have on-
Center approximate have an on- site medical
South, ly 3 hours site services.
Duval of school. Parents or
County, FL individual Youth guardians were
counseling attended responsible for
and 4 hours the local providing any
to 5 hours public needed medical
of group schools or services. The
counseling an most common
each week. alternative health problem
Family school was asthma.
counseling operated by
was the county.
provided on The shelter
a voluntary provided
basis. The independent
counselor- -living
to- skills
resident instruction
ratio was 1 and health
to 10. education.
Shelter Interface Female and Youth Youth The shelter did
Runaway male received enrolled in not have on-
Shelter, approximate the local site medical
Alachua ly 3 hours public services.
County, FL of schools Parents or
individual continued guardians were
counseling to attend responsible for
and 7 hours their providing any
of group regular needed medical
counseling classes. services. The
each week. Youth who most common
Family were not health problems
counseling enrolled in were colds,
was the local sinus
provided on schools infections, and
a voluntary attended an lice.
basis. The on-site
counselor- life skills
to- school that
resident was taught
ratio was 1 by a
to 8. retired
teacher.
The life
skills
school
included
remedial
instruction
,
independent
-living
skills
instruction
, and
health
education.
Shelter Coleman Female and Youth Youth On-site medical
House, male received residing in services
Fayette approximate the county included a
County, KY ly 2 hours attended nurse,
of local available 3
individual public days a week,
counseling schools. A and a doctor,
a week. teacher available 1 day
Group conducted a week. Youth
counseling on-site were given a
was not remedial physical within
provided. instruction 3 days after
The 4 hours a admittance to
counselor- week for the facility.
to- out-of- The most common
resident town youth health problems
ratio was 1 or youth were allergies
to 7. who had and colds.
dropped out
of school.
The shelter
also
offered
independent
-living
skills
instruction
and general
equivalency
diploma
preparation
.
Shelter Second Mile Female and Youth Youth The shelter did
Runaway male received either not have on-
Home, approximate attended site medical
Prince ly 4 hours the local services.
Georges of public Parents or
County, MD individual schools or guardians were
counseling received responsible for
and 4 hours tutoring at providing any
of group the needed medical
counseling facility. A care. The most
each week. licensed common health
The teacher problem for
counselor- provided both females
to- instruction and males was
resident every- hepatitis B.
ratio was 1 other-day The females
to 2. to youth commonly needed
not prenatal care
attending or treatment
the local for sexually
public transmitted
schools. diseases. The
Facility males commonly
staff needed
provided treatment for
instruction colds or dental
on the days problems.
the tutor
was not
available.
The shelter
also
provided
independent
-living
skills
instruction
, health
education,
and
parenting
classes.
Shelter Walden Female and Youth Youth who The shelter did
Sierra, male received 6 resided in not have on-
Inc., St. hours of the county site medical
Mary's individual attended services. The
County, MD counseling the local residents'
and 4 hours public health-care
of group schools. A needs were met
counseling licensed by local
each week. teacher health-care
The provided providers,
counselor- instruction including a
to- 4 hours a hospital,
resident day, 3 days women's clinic,
ratio was 1 a week to and pharmacy.
to 2. youth not Females
enrolled in commonly
the local requested
public gynecological
schools. services.
The shelter
also
provided
independent
-living
skills
instruction
, health
education,
and
parenting
classes.
Shelter Letot Female and Youth Youth The shelter did
Center's male received 6 attended not have on-
Emergency hours of the on- site medical
Shelter, individual site services. Youth
Dallas counseling school, received a
County, TX and 14 which had physical within
hours of two 2 days after
group classrooms arriving at the
counseling and two county's
each week. licensed juvenile
The teachers. detention
counselor- Daily co- center, which
to- educational also provided
resident classes any additional
ratio was 1 followed a medical care.
to 8. basic The most common
curriculum. health problems
The shelter were asthma,
also lice, scabies,
provided and sexually
independent transmitted
-living diseases.
skills
instruction
, drug and
health
education,
and
parenting
classes.
Group home Bardstown Male Youth The shelter The home did
Group Home received 1 did not not have on-
for Boys, hour of have an on- site medical
Jefferson individual site services. Youth
County, KY counseling school. received a
and 3 hours Youth physical from
of group attended an community
counseling off-site health-care
each week. alternative providers
The school within 7 days
counselor- operated by after
to- the state admittance to
resident and the the facility.
ratio was 1 local The most common
to 8. public health problems
school were severe
district. acne and dental
Daily co- problems.
educational
classes
followed a
basic
curriculum,
and most of
the classes
were
remedial.
The
alternative
school also
provided
counseling,
health
education,
and
recreationa
l
activities.
Group home Project Female Youth The group The home did
Respect received 4 home did not have on-
Group Home hours of not have an site medical
for Girls, individual on-site services.
Fayette counseling school. Physical and
County, KY and 3 hours Youth general medical
to 5 hours attended an care were
of group off-site obtained from
counseling alternative community
each week. school health-care
The operated by providers. The
counselor- the state most common
to- and local health problems
resident public were colds,
ratio was 1 school dental
to 4. district. problems,
Daily sexually
classes transmitted
followed a diseases, and
basic urinary tract
curriculum, infections.
and most of
the classes
were
remedial.
The
alternative
school also
provided
counseling,
health
education,
and
recreationa
l
activities.
Group home Salvation Female Youth Youth The home did
Army received at attended not have on-
Adolescent least 1 the local site medical
Treatment hour of public services.
Center, individual schools or Physicals and
Bexar counseling an on-site general medical
County, TX and 3 hours alternative care were
of group school. The obtained from
counseling alternative community
each week. school, health-care
The which providers. The
counselor- followed a most commonly
to- basic requested
resident curriculum, treatment needs
ratio was 1 had one were dental and
to 5. licensed gynecological
teacher services.
provided by
the local
school
district.
Group home San Antonio Female Youth The group The home did
Youth received 2 home did not have on-
Residential hours of not have an site medical
for individual on-site services. Youth
Females, counseling school. had to receive
Bexar each week. Youth a physical
County, TX No formal attended prior to
group the local admission and
counseling public general medical
was schools 5 care from
provided. days a community
The week. health-care
counselor- providers. The
to- most common
resident health problem
ratio was 1 was sprained
to 4. ankles.
Non- Practical Female Each youth The Center The home did
residential and was is an not have on-
program Cultural assigned an alternative site medical
Education adviser who school that services.
Center for spent helps girls Parents,
Girls, approximate obtain high guardians, or
Duval ly 5 hours school the students
County, FL a week credits or themselves were
discussing their responsible for
personal general obtaining any
and equivalency needed medical
academic diplomas. services.
issues and The program Prenatal care
goals. The had seven and
adviser- teachers gynecological
to- and seven services were
student classrooms. obtained from
ratio was 1 The daily community
to 10. In classes health-care
addition, a followed a providers with
therapist basic the permission
provided curriculum, of the parents
each girl and some of or guardians.
with at the classes The most common
least 1 were health problem
hour of remedial. was asthma.
counseling The program
(individual also
, group, or offered
family) independent
each week. -living
skills
instruction
, drug and
health
education,
and
parenting
classes.
--------------------------------------------------------------------------------
Source: Developed by GAO from information provided by facility
officials.
(See figure in printed edition.)Appendix V
NATIONAL SURVEY OF COUNTY JUVENILE
PROBATION DEPARTMENT OFFICIALS
========================================================== Appendix IV
(See figure in printed edition.)
(See figure in printed edition.)
(See figure in printed edition.)
(See figure in printed edition.)
(See figure in printed edition.)
(See figure in printed edition.)
(See figure in printed edition.)
(See figure in printed edition.)
MAJOR CONTRIBUTORS TO THIS REPORT
========================================================== Appendix VI
GENERAL GOVERNMENT DIVISION,
WASHINGTON, D.C.
James M. Blume, Assistant Director, Administration of Justice
Issues
Barry J. Seltser, Assistant Director
Thomas L. Davies, Senior Evaluator
William J. Sabol, Senior Social Science Analyst
Barbara A. Stolz, Senior Social Science Analyst
Joanne M. Parker, Senior Social Science Analyst
Brenda I. Rabinowitz, Evaluator
Maria D. Strudwick, Evaluator
Pamela V. Williams, Communications Analyst
Michelle D. Wiggins, Typist
OFFICE OF THE GENERAL COUNSEL,
WASHINGTON, D.C.
Jan B. Montgomery, Assistant General Counsel
DALLAS REGIONAL OFFICE
Danny R. Burton, Regional Management Representative
Teresa R. Russell, Evaluator-in-Charge
Christina M. Nicoloff, Senior Evaluator
Donna B. Svoboda, Evaluator
Frederick T. Lyles, Jr., Evaluator
Virginia B. Dandy, Technical Information Specialist
*** End of document. ***