Prosecution of Child Abuse A Meta-Analysis of Rates of Criminal Justice Decisions

Abstract

This study meta-analyzed rates of criminal justice decisions in 21 studies of prosecution of child abuse. Rates of referral to prosecution, filing charges, and incarceration varied considerably. Rates of carrying cases forward without dismissal were consistently 72% or greater. For cases carried forward, plea rates averaged 82% and conviction rates 94%. Compared to national data, child abuse was less likely to lead to filing charges and incarceration than most other felonies but more likely to be carried forward without dismissal. Diversion, guilty plea, and trial and conviction rates were about the same for child abuse and all violent crimes. Thus, prosecuting child abuse is generally neither feckless nor reckless. Rates can be misleading and cannot be the sole measure of prosecution success.

Full text

http://tva.sagepub.com
Trauma, Violence, & Abuse
DOI: 10.1177/1524838003256561
2003; 4; 323 Trauma Violence Abuse
Theodore P. Cross, Wendy A. Walsh, Monique Simone and Lisa M. Jones
Prosecution of Child Abuse: A Meta-Analysis of Rates of Criminal Justice Decisions
http://tva.sagepub.com/cgi/content/abstract/4/4/323
The online version of this article can be found at:
Published by:
http://www.sagepublications.com
can be found at:Trauma, Violence, & Abuse Additional services and information for
http://tva.sagepub.com/cgi/alerts Email Alerts:
http://tva.sagepub.com/subscriptions Subscriptions:
http://www.sagepub.com/journalsReprints.navReprints:
http://www.sagepub.com/journalsPermissions.navPermissions:
http://tva.sagepub.com/cgi/content/refs/4/4/323 Citations
at SAGE Publications - Full-Text Collections on February 9, 2009 http://tva.sagepub.comDownloaded from

10.1177/1524838003256561 ARTICLETRAUMA, VIOLENCE, & ABUSE / October 2003Cross et al. / PROSECUTION OF CHILD ABUSE
PROSECUTION OF CHILD ABUSE
A Meta-Analysis of Rates of Criminal Justice Decisions
THEODORE P. CROSS
WENDY A. WALSH
MONIQUE SIMONE
LISA M. JONES
University of New Hampshire
This study meta-analyzed rates of criminal justice decisions in 21 studies of prose-
cution of child abuse. Rates of referral to prosecution, filing charges, and incarcera-
tion varied considerably. Rates of carrying cases forward without dismissal were
consistently 72% or greater. For cases carried forward, plea rates averaged 82%
and conviction rates 94%. Compared to national data, child abuse was less likely to
lead to filing charges and incarceration than most other felonies but more likely to
be carried forward without dismissal. Diversion, guilty plea, and trial and convic-
tion rates were about the same for child abuse and all violent crimes. Thus, prose-
cuting child abuse is generally neither feckless nor reckless. Rates can be
misleading and cannot be the sole measure of prosecution success.
Key words: child abuse, child sexual abuse, prosecution, criminal justice, meta-analysis
HEARING ABOUT THE ABUSE OF A CHILD
provokes a visceral wish to punish the offender,
and most professionals agree that prosecution
and punishment of child maltreatment can be
an appropriate response (see Myers, 1994). Al-
though hard data are elusive, evaluation studies
and prosecutor surveys suggest a substantial in-
crease in the prosecution of child abuse during
the 1980s and 1990s, and many district attor-
ney’s offices around the country have estab-
lished specialized child abuse units (Smith &
Goretsky-Elstein, 1994). A 1994 survey of prose-
cutors found that 91% cited domestic violence
and child abuse as a source of workload prob-
lems in their offices (McEwen, 1995). Prosecu-
tion is our most consequential intervention in
response to child maltreatment, with a major
impact on perpetrators, victims, their families,
and the community.
Sometimes extraordinary cases have cap-
tured national attention, and prosecution of
child abuse has been portrayed as feckless and
323
AUTHORS’ NOTE: Grant No. 1999-JP-FX-1001 from the Office of Juvenile Justice and Delinquency Prevention supported this research.
The authors would like to thank the Family Violence Research Seminar, Janet Fine, and an anonymous reviewer for their reviews. They
also would like to thank Danielle LeBlanc, Katie Limoges, and Matt Palumbo for their assistance with the article. Correspondence regard
-
ing this article should be sent to Theodore P. Cross, Crimes Against Children Research Center, University of New Hampshire, 20 College
Road, #126, Horton Social Science Center, Durham, NH 03824; e-mail: ted.cross@unh.edu.
TRAUMA, VIOLENCE, & ABUSE, Vol. 4, No. 4, October 2003 323-340
DOI: 10.1177/1524838003256561
© 2003 Sage Publications
at SAGE Publications - Full-Text Collections on February 9, 2009 http://tva.sagepub.comDownloaded from

reckless. Stories about
prosecutors of child abuse
being unable to win con-
victions suggest that it
can be feckless (see
Dziech & Schudson,
1989), whereas stories
about overzealous prose-
cutors pursuing question-
able allegations suggest
that it can be reckless (see,
e.g., Wright, 1995). But
public and professional
understanding of prose-
cution of child abuse
should be based on data,
rather than newsworthy
cases that may not be rep-
resentative. Better data
would also help illumi-
nate the experience of
child victims in the
system.
Despite their impor-
tance, however, basic data
about prosecution of child
abuse have not been systematically compiled.
The federal Bureau of Justice Statistics reports
data on prosecution of many different crimes
(see, e.g., Reaves, 2001), including many that
would include child abuse (e.g., rape, assault),
but it does not generally present data separately
for victims of different ages.
A number of studies have reported rates for
several different decisions related to the prose-
cution of child abuse. This study analyzes these
rates for child abuse cases to address the follow-
ing questions:
• How often are substantiated or founded cases re-
ferred for prosecution?
• How often are charges filed?
• How often are cases diverted from prosecution to
treatment or other interventions?
• Once charges are filed, how often are cases carried
forward versus diverted or dismissed?
•
How often do child abuse defendants plead guilty
versus go to trial?
•
How often are child abuse defendants convicted?
•
How often are convicted defendants incarcerated?
This information is then combined to depict
case flow. Unlike previous reviews (e.g., Cross,
Whitcomb, & DeVos, 1995; Stroud, Martens, &
Barker, 2000), we include a comprehensive sam-
ple of studies and calculate aggregate statistics
using meta-analytic techniques. Next, we com-
pare rates for child abuse cases to overall rates
for felonies, violent crimes, and rape and sexual
assault charges to see how prosecution of child
abuse differs from prosecution of other crimes.
324 TRAUMA, VIOLENCE, & ABUSE / October 2003
KEY POINTS OF THE
RESEARCH REVIEW
•
Referral rates varied considerably across 4 stud
-
ies, ranging from 40% to 85%.
•
Charging rates varied considerably across 14
studies, ranging from 28% to 94%.
• Diversion rates ranged from 0% to 7% for studies
that reported disposition.
• The great majority of cases that were charged
were carried forward without dismissal or trans-
fer, a mean of 79% across 18 studies.
• Plea rates of cases carried forward were generally
high across 19 studies, with a mean of 82%,
whereas trial rates were correspondingly low,
with a mean of 18%.
• Conviction rates of cases carried forward were
consistently high, averaging 94% across 19
studies.
• Incarceration rates of those convicted varied con-
siderably, ranging from 24% to 96% across 14
studies.
• Child abuse cases were less likely to have charges
filed than most other felonies, were more likely to
be carried forward without dismissal, and had
lower incarceration rates.
• The only difference between rates in child abuse
cases and those of rapes and sexual assaults in
general was a significantlyhigher carried-forward
rate and a lower incarceration rate for child abuse
cases.
• On average, across the results reviewed, for a hy-
pothetical 100 alleged perpetrators referred to
prosecutors:
• 66 would be charged;
• 43 would plead guilty;
• 12 would be dismissed or transferred;
• 2 would be diverted to treatment;
• 6 would be convicted at trial;
• 3 would be acquitted at trial;
• 26 would be incarcerated.
Although hard data
are elusive,
evaluation studies
and prosecutor
surveys suggest a
substantial increase
in the prosecution of
child abuse during
the 1980s and 1990s,
and many district
attorney’s offices
around the country
have established
specialized child
abuse units (Smith &
Goretsky-Elstein,
1994). A 1994 survey
of prosecutors found
that 91% cited
domestic violence
and child abuse as a
source of workload
problems in their
offices (McEwen,
1995).
at SAGE Publications - Full-Text Collections on February 9, 2009 http://tva.sagepub.comDownloaded from

Finally, we generate a hypothetical case flow for
100 child abuse cases referred to prosecutors.
Special Nature of
Prosecution of Child Abuse
Prosecution of child abuse differs from most
prosecution in many ways (see, e.g., Davidson,
1997; Whitcomb, 1992), several of which may in-
fluence the rates analyzed here. First, the pro-
cess of investigation and referral to prosecutors
is different, which is likely to affect the popula-
tion of cases that are referred. Two different in-
vestigative agencies, police and child protective
services (CPS), regularly refer child abuse cases
to prosecutors, whereas only the police refer
most other crimes. One or both of these agencies
will be involved in all official child abuse inves-
tigations, either because private citizens contact
them directly or professionals such as health
care providers and school personnel contact
them following detection, disclosure, or suspi-
cion of abuse. CPS agencies’ primary mission is
to assess the child’s risk and need for protective
services.
The complicated, multidisciplinary nature of
investigating child abuse and the need to pro-
tect children in the process has led in part to the
development of specialized organizations to
carry out these functions. These range from spe-
cialized child abuse units within district attor-
neys’ offices, to multidisciplinary teams across
investigative agencies, to comprehensive
Children’s Advocacy Centers (CACs) (see
Sheppard & Zangrillo, 1996). These agencies
aim to improve the process of criminal justice in-
vestigations for children by making the experi-
ence more child-friendly and by coordinating
the efforts of the different investigators. CACs
have increased ninefold since 1994 and are now
established in 49 states across the United States
(National Children’s Alliance, 2001). The in-
volvement of CACs or other multidisciplinary
investigation organizations could have an im-
pact on the rates analyzed here.
Second, child abuse presents special chal-
lenges that can make prosecution difficult. It can
be very difficult to ascertain whether a criminal
act has occurred and who is culpable. In sexual
abuse cases particularly, evidence can be sparse.
Skilled prosecutors can use children’s state-
ments to find and present corroborative evi-
dence, but often prosecution relies heavily on
child victims’ testimony. Prosecution often de-
pends on families’ commitment to prosecution
and child victims’ ability as witnesses, credibil-
ity with juries, and capacity to withstand the
stress of a criminal trial (see, e.g., Davidson,
1997; McGregor, 1990; Whitcomb, 1992). Prose-
cution of both sexual and physical abuse also
depends on judges’ and juries’ willingness to
acknowledge a problem that historically has
been disbelieved, ignored, or discounted. The
special challenges of prosecuting child abuse
could influence criminal justice decision rates
through much of the process, including the de-
fendant’s decision to plead guilty versus go to
trial.
Third, child maltreatment may differ from
other crimes in the state’s interest in prosecution
and punishment. Professional opinion about
prosecution of child abuse has historically been
divided (see Harshbarger, 1987; Levesque, 1995;
Newberger, 1987; Peters, Dinsmore, & Toth,
1989), and prosecution is sometimes controver-
sial. Prosecution may be declined in favor of
therapeutic approaches, family or juvenile
court proceedings, or both (see Whitcomb &
Hardin, 1996 on coordination of criminal and
civil courts). Even when prosecutors believe a
criminal justice response is justified, they need
to weigh the costs and benefits of prosecution
for the victim, family, and community. Such
costs include the stress on children who may
need to testify and the emotional pain of seeing
an offending parent prosecuted, whereas bene-
fits include not only achieving justice but also
preventing further abuse by controlling offend-
ers. Such considerations could well influence
the rate at which cases are prosecuted and taken
to trial and the rate at which convicted defen-
dants are incarcerated.
How Prosecution of Child Abuse Works
In most ways, the prosecution of child abuse
parallels that of any prosecution (cf. Cross et al.,
1995; Davidson, 1997) (see Figure 1, adapted
Cross et al. / PROSECUTION OF CHILD ABUSE 325
at SAGE Publications - Full-Text Collections on February 9, 2009 http://tva.sagepub.comDownloaded from

from Boland, Mahanna, & Sones, 1992, for a dia-
gram of prosecution case flow). Referrals to
prosecution in police cases are relatively
straightforward, based on the evidence gath-
ered and the procedures of a particular jurisdic-
tion. For CPS cases, referral is discretionary for
many cases, although referral is mandatory for
certain types of more serious cases in most
jurisdictions (see, e.g., MacMurray, 1989; Men-
nerich, Martell, Cross, & White, 2002; Wilber,
1987).
Once prosecutors have received a referral of
child abuse from CPS or the police, they evalu-
ate the case to assess whether a crime has been
committed and whether they can prove it be-
yond a reasonable doubt without unduly hurt-
ing the victim. They may decline cases because
of insufficient evidence that a crime has been
committed or because the probability of obtain-
ing a conviction does not justify subjecting vic-
tims and their families to the stress of prosecu-
tion and possibly a trial (see, e.g., Smith &
Goretsky-Elstein, 1994; Whitcomb, 1992). De-
clined cases are not pursued further at that
point, although prosecutors may later file
charges if they obtain more evidence or if a
victim-witness who was initially unready to tes-
tify later becomes ready because of recovery or
maturation. Many declined cases are closed and
go no further in criminal court. They may be
pursued as lawsuits in civil court, and the alle-
gations may be considered in family, probate, or
juvenile court in relation to issues such as cus-
tody, visitation, and restraining orders.
Once prosecutors accept a case, they seek to
charge the defendants at grand jury or prelimi-
nary hearing and typically pursue conviction.
After prosecution is initiated, however, the case
may be dismissed by judges or dropped by
prosecutors if prosecution is no longer appro-
priate. Diversion is another alternative to prose-
cution when diversion programs are available
(see, e.g., Fridell, 1991). In diversion, prosecu-
tors agree to drop charges pending perpetra-
tors’ successful completion of an agreed-on
treatment program. If perpetrators do not fol-
low through with treatment, prosecutors can
then resume prosecution.
Once cases are carried forward without dis-
missal or diversion, defendants then must de-
cide whether to plead guilty or go to trial. Plea
bargaining between prosecutors and defen-
326 TRAUMA, VIOLENCE, & ABUSE / October 2003
Charged:
66
Carried
Forward:
52
Tried:
9
Guilty Plea:
43
Convicted:
49
Incarceration:
26
Referred for
Prosecution:
100
Acquitted:
3
Not charged:
34
Dismissed,
Transferred:
12
No incarceration:
23
Police
Referrals
CPS
Referrals
Diversion:
2
Final disposition of 100
hypothetical cases in order
of frequency:
Guilty plea 43
Not charged 34
Dismissed or transferred 12
Convicted at trial 6
Acquitted at trial 3
Diversion 2
100
Figure 1: Hypothetical Case Flow of 100 Child Abuse Cases Referred for Prosecution
at SAGE Publications - Full-Text Collections on February 9, 2009 http://tva.sagepub.comDownloaded from

dants may influence this decision. Defendants
may seek reduced sentences and prosecutors may
offer them in the interest of securing a guilty
plea. Defendants who go to trial may be found
guilty or acquitted, and those convicted through
a guilty plea or at trial will be sentenced.
METHOD
In contrast to a traditional literature review, a
meta-analysis statistically combines results in
an objective manner to summarize and synthe-
size selected empirical studies (Cooper &
Hedges, 1994; Lipsey & Wilson, 2001). In addi-
tion, meta-analysis provides a foundation for
future research by identifying what is and what
is not known about a particular topic.
Study Selection
Several strategies were used to identify po-
tential studies to be included in the analyses.
Electronic databases were searched for books,
published articles, dissertations, or conference
papers reporting empirical research on child
abuse prosecution. The searched databases in-
cluded PsychInfo, Lexus, Social Work Ab-
stracts, Social Sciences Abstracts, Educational
Resources Information Center (ERIC), and Dis-
sertation Abstracts International. Combinations
of the following keywords were used in the
searches: “prosecution,” “criminal justice,”
“courts,” “sentencing,” “child abuse,” and
“child maltreatment.” Efforts were also made to
identify relevant unpublished manuscripts or
articles in press by contacting child abuse re-
searchers through an electronic mail discussion
list (Child-Maltreatment-Research-L). Sources
were included in the meta-analysis if they pre-
sented quantitative data on child abuse prose-
cution case flow with information on at least one
of the following: referral rate, charging or prose-
cution acceptance rate, carried forward rate,
plea and trial rate, or incarceration rate. A total
of 21 studies were identified that met these crite-
ria. Most of the studies were relatively current,
with 5 studies published during the 1980s, 11
published during the 1990s, 4 completed during
the 2000s, and only 1 published before 1970 (see
Table 1) (these studies are also marked with an
asterisk in the references). The majority of stud-
ies (15) were from journals. The remaining
documents were monographs (1), books (1), fi-
nal reports (3), and a manuscript submitted for
publication (1). All of the studies were con-
ducted in the United States. Four additional
studies were identified but not included be-
cause 3 were conducted in Canada (Stevens &
Fischer, 1992; Trocme et al., 2002; Van Dam,
Halliday, & Bates, 1985), and 1 was conducted in
New South Wales, Australia (Cashmore &
Horsky, 1988).
Two of the 21 sources identified in the search
presented results from multiple samples or
multiple sites separately. Dolan (1985) reported
data from two communities. One community
implemented a special collaborative program
between the district attorney and CPS. In this
program, police-trained investigators con-
ducted special review of CPS cases with poten-
tial for prosecution. The other community was a
comparison using a standard CPS sample. A re-
search report from the American Bar Associa-
tion and the National Institute of Justice (Davis
& Wells, 1996) reported data separately for three
different sites: a child protection office, a sher-
iff’s office, and a prosecutor’s office. For pur-
poses of the meta-analysis, each of the different
samples presented in these sources was ana-
lyzed as a separate case. A total of 24 observa-
tions were included in the analysis from the
original 21 studies.
Of these 24, 19 (79%) concerned only sexual
abuse cases and 5 (21%) included a combination
of sexual and physical abuse cases. Data were
analyzed for the entire 24 and also separately for
the sexual abuse subsample and the sexual and
physical abuse subsample. For most outcomes,
differences between the two subsamples were
minimal. In addition, mean and median per-
centages for the sexual abuse subsample were
mostly within 5 percentage points of the total
sample. Therefore, the entire sample of 24 cases
was used.
In 54% of studies, samples were drawn from
prosecutors’ offices versus 25% from CPS or in-
vestigation agencies, 13% from police, and 8%
from human service agencies. The degree that
sample characteristics were available in these
studies varied greatly. Across 17 studies, 78% of
Cross et al. / PROSECUTION OF CHILD ABUSE 327
at SAGE Publications - Full-Text Collections on February 9, 2009 http://tva.sagepub.comDownloaded from

child victims were girls. Fourteen studies re
-
ported age of child victims, but some reported
means and others reported percentages of chil-
dren with age ranges. We calculated approxi-
mate means over the 14 studies, using exact
means when possible and estimating means us-
ing the information given when necessary.
Children across the 14 studies were on average
about 9 years old. In 6 studies that recorded
race, the overwhelming majority of child vic-
tims were White. Across 20 studies, 54% of the
cases involved intrafamilial abuse. In 11 studies
that reported age of offender, using the method
for approximating the mean described above,
offenders across the 11 studies were on average
about 34 years old.
Rates
Rates were calculated for the following deci-
sion points in the prosecution of child abuse: re-
ferral, filing criminal charges, diversion, carry-
ing cases forward (versus dismissing them),
guilty plea, trial, conviction, and incarceration.
In most studies, each perpetrator was the unit of
analysis and was considered an individual case,
although Mennerich et al. (2002) used the prose-
cutor’s entire case as the unit of analysis, even if
it included multiple perpetrators, as long as all
perpetrators had the same outcome. Each rate
was defined as follows:
Referral to prosecutors’ office. This is the pro-
portion of substantiated or founded cases re-
ferred to the prosecutors’ office by CPS or law
enforcement agencies. This rate equals the num-
ber of cases referred to the prosecutors’ office di-
vided by the number of cases substantiated or
founded by the source agency.
Charged. This is the proportion of referred
cases charged with a crime versus screened out.
The rate was calculated as the number of cases
in which criminal charges were filed divided by
the number of cases referred for prosecution.
Some studies used the term “accepted for prose-
cution” rather than “charged.”
Diversion. This is the proportion of charged
cases diverted from prosecution to treatment or
some other alternative. This was calculated as
the number of cases diverted divided by the
number of cases charged. For cases with an ab-
sence of diversion in disposition, the rate was
inferred to be zero, whether or not a diversion
program was available.
Carried forward. This is the proportion of
charged cases that go to trial or have guilty
pleas. Dismissed, diverted, transferred cases
were defined as not carried forward. The rate
was equal to the number of cases that resulted in
a guilty plea or trial divided by the number of
cases charged by the prosecutor’s office.
Guilty plea. This is the proportion of cases car-
ried forward that resulted in a guilty plea. The
rate was calculated as the number of cases with
a guilty plea divided by the number of cases that
were carried forward.
Trial. This is the proportion of cases carried
forward that led to a trial. The rate was calcu-
lated as the number of cases that went to trial di-
vided by the number of cases that were carried
forward. Note that the trial rate and the plea rate
total to 100% of all cases carried forward. Thus,
the trial rate equals one minus the plea rate,
and therefore results for these rates are essen-
tially two different ways of looking at the same
analysis.
Conviction. Conviction rate of cases carried
forward was calculated as the number of guilty
pleas and convictions at trial divided by the
number of cases that were carried forward.
Studies consistently had too few cases that went
to trial to calculate reliable rates of conviction at
trial.
Incarceration. This was the proportion of
convictions that resulted in incarcerations. The
rate was calculated as the number of cases that
led to incarceration divided by the number of
convictions.
Comparison Data
Comparison data were gathered from 12 na-
tional studies of prosecution in urban areas con-
ducted for the U.S. Department of Justice (DOJ)
328 TRAUMA, VIOLENCE, & ABUSE / October 2003
at SAGE Publications - Full-Text Collections on February 9, 2009 http://tva.sagepub.comDownloaded from

from 1977 to 1998. Each study gathered samples
of state court felony defendants for a 1-year time
frame in 30 to 40 of the most populous U.S.
counties. We calculated each rate whenever pos-
sible for every study. Comparison studies are
marked with a dagger (†) in the References sec-
tion (Boland & Brady, 1985; Boland, Brady,
Tyson, & Bassler, 1983; Boland, Conly,
Mahanna, Warner, & Sones, 1990; Boland,
Conly, Warner, Sones, & Martin, 1989; Boland,
Logan, Sones, & Martin, 1988; Brossi, 1979; Hart
& Reaves, 1999; Reaves, 1998; Reaves & Smith,
1995). Rates were calculated for three groups: (a)
all felony offenses, (b) all violent crimes, and (c)
all cases involving a charge of rape. Violent
crimes, as defined by the DOJ studies, included
murder, rape, robbery, assault, and such crimes
against children as child abuse, nonviolent or
nonforcible sexual assault, kidnapping, and
cruelty to a child. Note that each succeeding
comparison group was a subset of the previous
comparison group and that each comparison
group included child abuse cases—it was im-
possible to separate them from crimes against
adults. The comparison of rates from the federal
data with rates from the child abuse studies was
slightly conservative as a result.
Because some of these studies only reported
aggregate results across jurisdictions, the unit of
analysis for the comparison data was the entire
United States. Because the federal comparison
data were drawn primarily from urban areas
“that account for the vast majority of all re-
ported crimes” (approximately 85% according
to reports from the FBI) (Boland & Sones, 1986,
p. 3), these rates could reasonably be viewed as
national.
Analysis
Data were aggregated across studies for the
following rates: referral to prosecution, charg-
ing, diversion, carrying forward, pleading
guilty, going to trial (which is simply one minus
the guilty plea rate), conviction (of cases carried
forward), and incarceration. Most studies only
reported data on some of these rates, so the ac-
tual N of cases meta-analyzed varied across the
specific rates.
Mean rates for each de
-
cision point were calcu-
lated for the child abuse
studies. Studies were not
weighted by sample size
(see Light & Pillemer,
1984; Lipsey & Wilson,
2001), because the source
of the sample (CPS, po-
lice, special agencies) ap-
peared to be directly re-
lated to its size. We felt
that weighting the sample
would have given certain
jurisdictions of certain
types undue influence in
the results. A parallel analysis of weighted sta-
tistics yielded very similar results, however.
The weighted mean rates differed from their un-
weighted counterparts by no more than 4.0%.
We also calculated median rates, which differed
from means by no more than 6.5% and so were
not used further. For each rate, the Q statistic
(Lipsey & Wilson, 2001) was calculated to assess
whether the variation was likely due to system-
atic differences across the samples or instead
was better explained by chance (sampling
error).
Mean rates were also calculated for each of
the three categories of felonies in the compari-
son data. The mean rates for the child abuse
sample were then compared to the comparison
mean rates using one-sample t tests, which
treated the comparison mean rates as popula-
tion values. Note that it would have been inap-
propriate to use standard comparison of means
procedures (e.g., independent-sample t tests) to
contrast the child abuse studies and national
comparison studies because the unit of analysis
is jurisdiction for the child abuse studies but the
entire country for the federal comparison
studies.
Adapting the method of Boland and col-
leagues (e.g., Boland et al., 1992; see also Cross
et al., 1995), we also examined the hypothetical
flow of 100 child abuse cases through different
stages of prosecutorial decision making. This
model uses the mean rates identified in the
studies that were reviewed and provides a sce
-
nario representing “typical” prosecution out
-
Cross et al. / PROSECUTION OF CHILD ABUSE 329
Comparison data
were gathered from
12 national studies of
prosecution in urban
areas conducted for
the U.S. Department
of Justice (DOJ) from
1977 to 1998. Each
study gathered
samples of state court
felony defendants for
a 1-year time frame
in 30 to 40 of the most
populous U.S.
counties.
at SAGE Publications - Full-Text Collections on February 9, 2009 http://tva.sagepub.comDownloaded from

comes for a set of 100 cases at the point at which
they have been referred for prosecution. This
analysis suggests general trends, even though it
does not take into account the possibility that
rates at different points may be correlated nor
that mean rates may not be representative of
many jurisdictions.
RESULTS
Table 1 presents the rates for each of the child
abuse studies, and Figure 2 presents boxplots
summarizing the distribution for each rate in
the child abuse sample. Table 2 presents statis-
tics on each rate in both the child abuse sample
and comparison samples.
Rates
Referral to district attorney. The unweighted
mean referral rate was 56% across four studies,
but the mean was not useful because of the vari-
ability in rates. A special program that included
both health care and police intervention in child
abuse cases had a high rate of 85% (Rogers,
1982). Dolan’s (1985) quasi-experiment found
that a community that had a special CPS-DA
collaborative investigation program within CPS
referred 59% of cases to prosecutors. On the
other hand, the two standard CPS samples (in-
cluding Dolan’s comparison community) both
had a substantially lower referral rate of 40%.
Charging. Across the 13 studies that had
charging or prosecution rates, more than half of
child abuse cases considered by prosecutors
were accepted, but again this rate is so variable
that the mean is not useful. This rate ranged
from 28% to 94%, and even the middle half
(interquartile range) of studies varied by 33%.
The majority of studies (8) had charging rates
between 48% and 76%.
Diversion/deferred. For the 12 studies that re-
ported diversion or from which zero diversion
could be inferred from an absence of diversion
in disposition, an average of 2% of cases were
diverted. For the six studies in which diversion
was reported, an average of 3% of cases were di
-
verted in these studies. However, given that
most studies did not report diversion, it seems
likely that many jurisdictions do not use diver-
sion, at least formally, and the mean rate of for-
mal diversion is much lower. It is possible, of
course, that an informal type of diversion was
used in these jurisdictions, in which prosecutors
agree not to file charges if the alleged perpetra-
tor pursues treatment, but retain the option of
filing charges later. But we have no way of mea-
suring that in this meta-analysis.
Carrying cases forward. The carried forward
rate refers to the proportion of cases that, once
charged, are carried forward either to trial or to
a guilty plea. In other words, this represents the
proportion of charged cases that were not later
dismissed, diverted, or transferred. In 15 of the
18 studies for which carried-forward rates could
be calculated, the rate was 72% or greater. Two
outliers (Bradshaw & Marks, 1990; DeFrancis,
1969) with small carried-forward rates collected
some or all of their data from the 1960s and
1970s. These cases occurred before the increased
sophistication about sexual abuse and increased
coordination among police, prosecutors, and
other professionals that arose in the 1980s. A
third outlier (Dolan, 1985) had 50% of cases car-
ried forward in the community with a special
sexual abuse unit, but this study had a small N
of 30.
Pleading guilty. The plea rate in cases carried
forward was consistently high, exceeding 83%
in 14 of the 19 studies. In one outlier, DeJong and
Rose (1991), the authors noted that the strict sen-
tencing in that state for guilty pleas may have
led to the low rate (39%) of pleas. This state re-
quired those who pled to serve 4 years in jail,
and the mandatory minimum sentence for fel-
ony child sexual abuse was 5 to 10 years.
Martone, Jaudes, and Cavins (1996), in another
outlier, reported case disposition for alleged
perpetrators charged only with felonies, with
their plea rate of 57%. Felony cases of sexual
abuse or assault may be more likely to proceed
to trial versus plea rather than misdemeanors.
The DeFrancis (1969) study reported that its
most significant finding was the high dismissal
rate of prosecutions (44%), which may severely
impact the low number of pleas (61%), resulting
in a third outlier.
330 TRAUMA, VIOLENCE, & ABUSE / October 2003
at SAGE Publications - Full-Text Collections on February 9, 2009 http://tva.sagepub.comDownloaded from

TABLE 1: Data in the Meta-Analysis
Conviction
Abuse Carried Guilty of Carried
Study Sample
N
a
Type Referral Charged Diversion Forward Plea Trial Forward Incarceration
1. De Francis, 1969 Brooklyn Society for the 143 Sexual — — .— .43 .61 .39 .85 .43
Prevention of Cruelty to Children
2. Conte & Berliner, 1981 Prosecutors 84 Sexual — — .— .99 .77 .23 .91 .59
3. Rogers, 1982 Specialized health and police program 261 Sexual .85 .48 .00
b
.72 .87 .13 .98 —
4. Dolan, 1985
c
CPS 83 Sexual .59 .76 .00
b
.50 .94 .06 1.00 .25
5. Dolan, 1985
d
CPS 40 Sexual .40 .75 .00
b
.78 .71 .29 .86 .33
6. Chapman & Smith, 1987 Prosecutors 249 Sexual — .63 .06 .82 .88 .13 .94 .60
7. MacMurray, 1989 Prosecutors 87 Sexual — .49 .————— —
8. Bradshaw & Marks, 1990 Prosecutors 350 Sexual — .74 .00
b
.49 .83 .17 .96 .52
9. DeJong & Rose, 1991 Prosecutors 115 Sexual — — .— — .39 .61 .92 —
10. Goodman et al., 1992 Prosecutors 170 Sexual — — .— .87 .86 .14 .93 .59
11. Tjaden & Thoennes, 1992 CPS 34 Physical — — .— .79 .85 .15 .93 .24
and sexual
12. Gray, 1993 Prosecutors 303 Physical — — .— — .88 .12 — —
and sexual
13. Smith & Goretsky-Elstein, 1993 Prosecutors 919 Sexual — — .03 .87 .86 .14 .95 .59
14. Cross, Whitcomb, & De Vos, 1995 Prosecutors 552 Sexual — .60 .02 .92 .84 .16 .93 .78
15. Davis & Wells,1996
e
CPS 225 Physical
and sexual .40 .92 .01 .85 .94 .06 .99 —
16. Davis & Wells, 1996
f
Sheriff’s office 99 Physical
and sexual — .94 .06 .77 .95 .05 .97 —
17. Davis & Wells, 1996
g
Prosecutors 6,111 Sexual — .68 .00
b
.87 — — .99 .32
18. Martone, Jaudes,& Cavins, 1996 Hospital investigation team 77 Sexual — .86 .00
b
.88 .57 .43 .86 .90
19. Brewer, Rowe, & Brewer, 1997 Private investigative agency 200 Sexual — .39 .————— —
20. Cheit & Goldschmidt, 1997 Arraigned cases 1,138 Sexual — — .— .84 .87 .13 .95 .45
21. Cullen, Smith, Funk, & Haaf, 2000 Prosecutors 201 Sexual — — .— .89 — — — —
22. Faller & Henry, 2000 Criminal court files 263 Sexual — — .004 .91 .94 .06 .96 .96
23. Mennerich, Martell, Cross, Prosecutors 703 Physical — .28 .————— —
& White, 2002 and sexual
24. Stroud, Martens, & Barker, 2000 Children’s Advocacy Center 174 Sexual — — .— — .97 .03 .99 —
NOTE: CPS = child protective services.
a.
N
refers to the denominator for the first available rate.
b. Inferred from absence of diversion in disposition.
c. Experimental community.
d. Control community.
e.CPS case study.
f.Sheriff’s office case study.
g. Case study in one site for multiple years.
331
at SAGE Publications - Full-Text Collections on February 9, 2009 http://tva.sagepub.comDownloaded from

Going to trial. Note that the trial rate was sim-
ply one minus the plea rate, so these rates were
merely two different ways of expressing the
same phenomenon. Trial rates varied from 3%
to 61%; the highest rate was, of course, DeJong
and Rose (1991), with a mean of 18% (median =
14%) of carried-forward
cases proceeding to trial.
Fourteen of the 19 studies
with trial rates had rates
of 16% or less.
Conviction. Of those
cases carried forward and
not dismissed, diverted,
or transferred, a large per-
centage (mean = 94%)
ended in convictions ei-
ther through guilty pleas
or at trial. Sixteen of the 19
studies had rates 90% or
greater. Studies consis-
tently had too few cases
that went to trial to calcu-
late reliable rates of con-
viction at trial.
Incarceration. Incarceration rates of those con-
victed varied considerably. Rates for the 14
studies ranged from 24% to 96%, with a mean of
54%. Studies were fairly equally distributed
across the range. For example, 2 studies had in-
carceration rates of 25% or less, 2 had rates be-
tween 30% and 39%, 2 had rates between 40%
and 49%, 4 had rates between 50% and 59%, 2
had rates between 60% and 80%, and 2 had rates
of 90% or more.
Systematic Versus Chance
Variation in Rates
For the child abuse studies, the Q statistic was
statistically significant at p < .001 for all but one
rate in Table 2 (for diversion). This provides
evidence that the variation in rates is system-
atic, that is, due to real differences between
the studies analyzed and not just sampling
error or chance. It is beyond the scope of this
article to explore what jurisdictional, organi-
zational, or sample variables could explain
this variation, but this suggests the value of
additional meta-analytic research (see below).
332 TRAUMA, VIOLENCE, & ABUSE / October 2003
Of those cases
carried forward and
not dismissed,
diverted, or
transferred, a large
percentage (mean =
94%) ended in
convictions either
through guilty pleas
or at trial. Sixteen of
the 19 studies had
rates 90% or greater.
Studies consistently
had too few cases
that went to trial to
calculate reliable
rates of conviction at
trial.
141919191812134N =
Criminal Justice Rates
Incarceration
Conviction
Trial
Plea
Carried forward
Diversion
Charged
Referred
1.0
.8
.6
.4
.2
0.0
1
18
91
18
9
4
8
1
16
Figure 2: Boxplots of Rates for Criminal Justice Decisions
NOTE: Blacklines representthe median.Boxesrepresent the middlehalf ofthedistribution.Linesextendingfromboxes(“whiskers”)rep-
resent the remainder of the distribution except for outliers and extremes, which are represented by O and asterisk (*), respectively. The
study number for each outlier and extreme is shown.
at SAGE Publications - Full-Text Collections on February 9, 2009 http://tva.sagepub.comDownloaded from

Comparison to Other Crimes
Table 2 also presents results for the compari-
son of rates in child abuse cases to all felonies,
violent crimes, and rapes and sexual assaults in
general. Child abuse cases were less likely to
have charges filed than felonies overall and
other violent crimes, but the charging rate was
similar to the rate for rape and other sexual as-
sault generally. The diversion rate for each of the
comparison categories was similar to the 2%
rate of the child abuse studies—3% for all felo-
nies and 1% for violent offenses and rape and
other sexual assault. Child abuse cases were
substantially and significantly more likely to be
carried forward without dismissal than violent
crimes and rape and other sexual assault. Guilty
plea and trial rates for child abuse cases were
similar to all the comparison categories, and the
differences were not statistically significant.
Carried-forward child abuse cases were as
likely to lead to conviction as all three compari-
son categories. The incarceration rate for child
abuse cases was lower than all three comparison
categories and significantly lower than violent
crimes and rape and sexual assault.
Hypothetical Case Flow
Figure 1 presents case flow for a hypothetical
100 cases referred to prosecutors—this analysis
focused on cases after referral because there
were few studies on referral and their results
were too variable. This analysis is based on av-
eraging across studies. About 66 out of 100 cases
referred to prosecutors would be charged with a
crime. Fifty-two would be carried forward to
guilty plea or trial. Forty-nine would be con-
victed, 43 of those through a guilty plea.
Twenty-six would be incarcerated. The text box
on Figure 1 listing dispositions in order of fre-
quency shows that a substantial majority of the
100 defendants are either not charged or plead
guilty, and only 9 go to trial.
DISCUSSION
These data demonstrate that most substanti-
ated and founded child abuse cases do not lead
to prosecution but that the majority of cases that
are prosecuted do end in conviction, the great
majority by guilty plea. A summary of the find-
ings is presented in Key Points of the Research
Review. The small number of available studies
varied considerably in the proportion of cases
that were sent to prosecutors and the proportion
on which prosecutors filed charges. But there
was considerable consistency in carrying cases
forward, obtaining guilty pleas, and convic-
tions once cases were charged. Incarceration
rates were variable. Rates for child abuse cases
resembled rates for all felonies generally, except
Cross et al. / PROSECUTION OF CHILD ABUSE 333
TABLE 2:
Summary Table of Rates of Criminal Justice Action
Child
Abuse
Studies
Comparison Studies
Rape &
Other
All Violent Sexual
Felonies Offenses Assault
Number Number
Criminal Justice Action of Studies
MSD
Minimum Maximum of Studies
MMM
Referral to prosecution 4 .56 .21 .40 .85 8 — — —
Charged 13 .66 .20 .28 .94 12 .79* .79* .69
Diversion 12
a
.02 .02 .00 .07 12 .03 .01 .01
Carried forward 18 .79 .16 .43 .99 12 .73 .65** .69*
Guilty plea 19 .82 .15 .39 .97 12 .87 .81 .76
Trial 19 .18 .15 .03 .61 12 .13 .19 .24
Conviction 19 .94 .05 .85 1.00 12 .96 .94 .93
Incarceration 14 .54 .23 .24 .96 12 .66 .73** .74**
NOTE: Ratesinthechildabusestudiescamefromindividualjurisdictionsoragencies,andratesinthecomparisonstudieswerenational.
a. For the six studies in which diversion was reported,
M
= .03,
SD
= .03, Minimum = .004, Maximum = .07.
*
p
< .05. **
p
< .01.
at SAGE Publications - Full-Text Collections on February 9, 2009 http://tva.sagepub.comDownloaded from

that they were less likely to have charges filed.
Furthermore, child abuse cases were more
likely to be carried forward but less likely to
lead to incarceration than violent offenses and
rape and sexual assault.
Understanding Rates
Referral rate. This most fundamental step is
probably the least understood. Most studies of
prosecution of child abuse use samples of cases
that have already been referred to prosecutors
(see Stroud et al., 2000), limiting what can be
learned about the referral process itself. Referral
rates differed greatly by the type of agency. We
do not know to what extent variation in referral
stems from differences in policy and procedure
versus differences in decision making. We are
aware of no studies that look in detail at the pro-
cesses of referral in the prosecution of child
abuse, although Stroud et al. (2000) examine
variables that predict referral.
Charging rates. Charging rates vary greatly
across studies, although high percentages of
those actually charged are convicted. Thus, the
best predictor of whether alleged perpetrators
in a prosecution sample are punished is
whether or not they are charged (see Cross et al.,
1995). There is now a
moderate body of litera-
ture on rates at which
cases referred to prose-
cutors on child sexual
abuse allegations are
charged and some stud-
ies that included physi-
cal abuse as well.
Variation across juris-
dictions is the central
feature of these data,
however, and the mean
and median are not nec-
essarily representative.
We can think of several
possible explanations
for this variation.
First, differences in
the screening process
across jurisdictions may
lead to differences in charging rate, not because
criteria for case selection are so different but be-
cause the organizational procedures for screen-
ing cases are different and are not well-captured
by the charging rate. For example, screening in
one jurisdiction may be a two-stage process, and
the charging rate may only measure one of those
stages. It is possible, for example, that police or
CPS may consult with prosecutors about cases
in one jurisdiction prior to referral and infor-
mally screen out cases, whereas prosecutors in
another jurisdiction may screen cases only after
they have been referred for a charging decision.
Second, differences in charging rates proba-
bly reflect differences in the population of cases
across studies. Some of this is obvious from the
different nature of the population in these stud-
ies—cases seen by CPS, sheriff’s offices, and
hospital investigations teams, for example, are
likely to differ. But charging rates were not
linked in obvious ways to the setting from
which the sample came, and there was still con-
siderable variation within similar samples, for
example, cases drawn from prosecutors’ offices.
Because of local or state policy or practice, dif-
ferent communities may apply lower or higher
thresholds for referral for prosecution. States
mandate referral from CPS to law enforcement
or prosecutors for certain categories of abuse,
but these laws vary across states. Variations in
training of police, prosecutors, and judges may
create differences in criteria for investigating
cases. Even prosecutors applying identical crite-
ria in two different communities may achieve
different prosecution rates because of differ-
ences in the potential for prosecution of the
cases sent to them. In addition, communities
may diverge in the incidence of child abuse of a
type and severity that warrants prosecution or
in the willingness of child victims and their fam-
ilies to support prosecution.
Third, some variation may be due to disparity
among district attorneys in their commitment to
and investment in prosecuting child abuse.
Some district attorneys may have a broader per-
spective on the role of prosecution, and believe
in prosecuting a wider range of cases than oth-
ers. They may vary as well in the weight they
give to victims’ wishes for or against prosecu
-
334 TRAUMA, VIOLENCE, & ABUSE / October 2003
Because of local or
state policy or
practice, different
communities may
apply lower or higher
thresholds for referral
for prosecution. States
mandate referral from
CPS to law
enforcement or
prosecutors for certain
categories of abuse,
but these laws vary
across states.
Variations in training of
police, prosecutors,
and judges may
create differences in
criteria for
investigating cases.
at SAGE Publications - Full-Text Collections on February 9, 2009 http://tva.sagepub.comDownloaded from

tion. Training may play an important role here,
too.
Fourth, district attorneys are likely to vary in
the priority they give prosecuting child abuse,
even when they agree in principle. DAs must re-
spond to a wide range of criminal behavior.
Prosecution of child abuse requires special com-
mitment because of its many special demands,
such as interviewing children, supporting fami-
lies, and coordinating with other professionals.
Some district attorney’s offices have invested
resources by hiring specialized professionals,
developing specialized units, and participating
in multidisciplinary teams and agencies like
CACs. Many district attorneys’ offices have not
made this commitment, however, and may be
less likely to charge cases.
Note that increased zeal and resources for
prosecuting child abuse could paradoxically
lower charging rates by influencing which cases
are referred to prosecutors. If a community un-
derstands that prosecutors have an increased
willingness, ability, or both to pursue child
abuse cases, it may refer more cases that are dif-
ficult to investigate and prosecute. Committed
prosecutors may encourage these referrals and
screen in cases that other prosecutors would re-
ject. Prosecutors may, for example, be willing to
interview younger children and look for corrob-
orative evidence on allegations that other offices
would not find sufficiently credible to spend
time on. Many of these cases may be summarily
closed without taking much time or having
much impact on the criminal justice system, but
they would have an effect on the charging rate.
Fifth, prosecutors’ offices may vary in their
skill in developing cases that can be prosecuted.
Prosecution of child abuse is a specialized en-
deavor that requires skill in and commitment to
child interviewing, gathering corroborative evi-
dence, and many other elements of investiga-
tion. It is likely that this skill is not evenly dis-
tributed, so some prosecutors’ offices are likely
to be better than others at developing cases that
lead to charges.
Diversion. The small number of studies in
which diversion is reported and the small per-
centages of diverted cases even in those studies
suggest that diversion is not used frequently
and rarely programmatically or systematically.
Time and resources may be lacking to develop
formal diversion programs, and the risks of
reoffending may often seem to outweigh the
benefits of diversion. An informal type of diver-
sion may be used even when formal diversion is
lacking. Rather than formally charging alleged
perpetrators with a crime and then dropping
the charges as part of a diversion agreement,
prosecutors may agree informally not to charge
if alleged perpetrators pursue treatment, leav-
ing open the option of charging later. The rarity
of diversion programs parallels the rarity with
which it is discussed in the professional litera-
ture. Is diversion an alternative that has not
been explored sufficiently?
Carried forward.This is the first of three rates
examined here that demonstrated considerable
consistency across studies. This consistency
suggests that prosecutors are charging cases
that they judge to have high probabilities of con-
viction. The high rates at which cases were car-
ried forward without dismissal may reflect their
seriousness and the care with which prosecu-
tors decide to file charges, which may in turn be
related to the investment of time and resources
they require.
Plea and conviction. Echoing Cross et al. (1995),
one of the most important findings from this re-
search is that child abuse trials are relatively
rare, certainly as a proportion of cases investi-
gated but even as a proportion of cases charged.
The rates at which defendants pled guilty ver-
sus went to trial were consistently high. The
consistently high rates suggest that prosecutors
in many different jurisdictions have been able to
select, prosecute, and plea-bargain effectively
enough to obtain guilty pleas in a large majority
of cases that they carry forward. The large pro-
portion of guilty pleas of those cases carried for-
ward ensures high overall conviction rates. The
statistics for prosecution of child abuse thus re-
semble those of felony arrests overall, which
also feature high plea and conviction rates for
cases carried forward.
Note that the consistency in plea rates and
conviction rates does not necessarily mean that
crimes on which offenders were sentenced were
equally severe across jurisdictions. Prosecutors
Cross et al. / PROSECUTION OF CHILD ABUSE 335
at SAGE Publications - Full-Text Collections on February 9, 2009 http://tva.sagepub.comDownloaded from

are likely making decisions to charge and carry
forward cases based on their offices’ ability to
obtain convictions in a given case. One office
may obtain a high conviction rate but only pur-
sue the most severe cases, whereas another of-
fice may have an identical conviction rate but
pursue a wider range of cases.
Incarceration. The great variation in incarcera-
tion rates suggests that there is no consistently
applied standard regard-
ing sentencing of child
abuse across jurisdic-
tions. Cheit and
Goldschmidt (1997) ex-
pressed concern that sen-
tences for child sex
offenders in Rhode Island
were more lenient than
those for other felonies of
comparable severity,
whereas Champion (1988)
found that sentences for
child sexual abusers were
more severe. Given the
considerable variation in
sentencing found here,
the difference between
these two studies may
simply be one manifesta-
tion of more widespread
variation in the difference
between sentences for
child sexual offenses and
for other crimes.
Differences in standards for punishment
could be one explanation for the wide variation
in incarceration rates. Judges and juries across
jurisdictions may differ on their criteria for
choosing to incarcerate, and states may vary in
their sentencing standards. Public and judicial
opinion about appropriate sentences may vary
across states and regions. In addition, effective
alternatives to incarceration may promote
lower incarceration in some jurisdictions, but
the lack thereof may lead to higher incarceration
rates in other jurisdictions. Variation in incar-
ceration may stem from differences in the popu-
lation of cases referred for prosecution. Regard
-
less of their cause, these results suggest that
standards for incarceration of defendants con
-
victed of child abuse crimes need further
examination. Note that the mere fact of being in-
carcerated may be less important than the
length of incarceration, which was beyond our
scope but should be the focus of new research.
Comparison to other crimes. Child abuse and
other violent crimes had very similar rates. This
suggests that there is no radical difference be-
tween the prosecution of child abuse and other
violent crimes, particularly rape and sexual as-
sault cases in general. Perhaps the most impor-
tant difference is that prosecutors accept lower
proportions of child abuse cases, but they also
appear to dismiss fewer as well.
From these data, it is impossible to tell why
the charging rate was lower for child abuse than
for other violent crimes. To what extent is it be-
cause allegations were false, insufficient evi-
dence was found to initiate prosecution, or pros-
ecution was rejected in the interest of the child,
family, or community? The higher rate of carry-
ing child abuse cases forward without dismissal
may stem from careful screening of child abuse
cases in the charging process. The lower incar-
ceration rate may be a function of child abuse
perpetrators’ frequent close relationship to vic-
tims, which may lead judges to perceive child
abuse as less serious or to avoid incarceration
because of the impact they believe it will have
on the family. Further research is needed here.
Implications for Practice,
Policy, and Research
The results here suggest that using rates to
define prosecution success could be misleading.
On one hand, carried-forward and conviction
rates appear to be so consistent that they may be
insensitive to differences in effectiveness. On
the other hand, charging rates are so variable
and potentially subject to differences in mea-
surement, referral, screening practices, and
populations that any contrast in rates is hard to
interpret. Good programs willing to evaluate
more difficult cases might be “punished” with
lower charging rates. Charging and conviction
rates can seem persuasive, but any conclusion
drawn from them should be examined skepti-
cally, and they should never be the sole measure
of an agency’s success.
336 TRAUMA, VIOLENCE, & ABUSE / October 2003
The higher rate of
carrying child abuse
cases forward without
dismissal may stem
from careful
screening of child
abuse cases in the
charging process.
The lower
incarceration rate
may be a function of
child abuse
perpetrators’ frequent
close relationship to
victims, which may
lead judges to
perceive child abuse
as less serious or to
avoid incarceration
because of the
impact they believe it
will have on the
family.
at SAGE Publications - Full-Text Collections on February 9, 2009 http://tva.sagepub.comDownloaded from

Note further that we have used a consistent
definition of rates here, but other parties that are
reporting rates may calculate them in different
ways. Charging rates are very different depend-
ing on the denominator used to calculate them
(the numbers investigated, substantiated, re-
ferred, etc.). The same is true of conviction rates.
Variation in defining and calculating rates adds
to the confusion and misinterpretation.
Much public and professional attention is de-
voted to trials of child abuse cases, but the num-
ber of cases in which alleged perpetrators are
not charged or defendants plead guilty is more
than 8 times greater than those going to trial. We
know much less about what happens after in-
vestigation and prosecution for the perpetrator,
the child, and the family in these cases, nor have
models of intervention been described for these
situations.
Limitations
The relatively small number of studies avail-
able, especially for some decisions, limits this re-
search. Several studies have small sample sizes
as well, making rates subject to sampling error.
Furthermore, some of the agencies studied were
either selected for research because of their spe-
cial characteristics (e.g., Dolan, 1985) or agreed
to participate in research because of their special
interest in prosecuting child abuse (e.g., Cross
et al., 1995). This may limit the generalizability
of these results. The fact that the comparison
samples included child abuse cases, although
presumably a small number, may obscure dif-
ferences between child abuse cases and other
crimes to some degree.
Future Research
We see three directions for new research.
First, more research on rates needs to be done.
Better reporting and dissemination of data from
district attorney and child protective services
offices could increase the number of studies ten-
fold and also better represent standard practice.
Because decisions at different stages may be in-
terdependent, studies should report case flow
as a whole rather than individual rates. New
studies should report rates within jurisdictions
for multiple years and tie variation to changes in
policy and practice, such as the introduction of
victim witness advocates, multidisciplinary
teams, or CACs.
Second, more could be done meta-analyzing
studies of prosecution rates, especially given
the systematic variation revealed here by the Q
statistic. Examining how rates are correlated
with the year of the study (see Light & Pillemer,
1984) could help identify the impact of changes
in public knowledge and opinion or the devel-
opment of innovations such as multidisci-
plinary investigation teams, mandatory referral
from CPS to law enforcement, and sex offender
registries. Additional meta-analyses could ex-
plore how rates vary by jurisdictional or organi-
zational factors or by population characteris-
tics. Studies have examined individual case
factors related to the decision to prosecute (see,
e.g., Cross, DeVos, & Whitcomb, 1994), but we
know little about what leads agencies to vary on
prosecution. The variation seen in the few stud-
ies on referral and sentencing also needs more
research.
Third, research must go beyond rates. Many
studies have examined a variety of aspects of
prosecution, such as time span of case process-
ing (see Ostrom & Hanson, 2000), plea bargain-
ing (see, e.g., Nardulli, Eisenstein, & Flemming,
1988) and sentencing, but few such studies fo-
cus on child maltreatment. Studies should also
link case flow data to survey or interview data
regarding prosecutors’ decision making on in-
dividual cases to better understand why cases
flow along certain paths. Alternative indicators
of prosecutor success should be developed,
such as measures of the quality of the investiga-
tion and preparation process. Given the
multidisciplinary response to child maltreat-
ment, research should also examine prosecution
in relation to agencies such as CPS and civil
courts. Research should examine what happens
when cases are not prosecuted, including child
protective action, civil court action, and thera-
peutic intervention. Cross, Martell, McDonald,
and Ahl (1999) found that children were more
likely to be placed outside the home in cases that
were declined by prosecutors. Martell (2002)
did not find this effect in a replication, but pro
-
tective social workers did report that various
Cross et al. / PROSECUTION OF CHILD ABUSE 337
at SAGE Publications - Full-Text Collections on February 9, 2009 http://tva.sagepub.comDownloaded from

prosecutor actions had been helpful to avoid
child placement. Research should also examine
what happens for children, families, and defen-
dants following guilty pleas.
CONCLUSION
Much more research is needed, but even
these few studies provide core knowledge that
can inform policy and practice. They suggest
the value of a proportionate view of prosecution
of child abuse. In this view, prosecution is a ma-
jor response but still applies to a minority of
cases, mostly sexual abuse. On the whole, pros-
ecution of child abuse is not feckless, in that
most cases that are charged end in conviction.
Neither is it reckless, in that case flow resembles
that of other felony cases and trials are relatively
few. Cases without charges filed and cases that
end in guilty pleas should receive attention
more commensurate with their numbers.
Evaluation of prosecution must move be-
yond prosecution rates and conviction rates,
which must be interpreted with healthy skepti-
cism. Alternative measures of prosecution suc-
cess must be developed. On this sometimes
overheated topic, data can be a tool for rational
analysis.
338 TRAUMA, VIOLENCE, & ABUSE / October 2003
IMPLICATIONS FOR PRACTICE, POLICY, AND RESEARCH
• We should not view prosecution of child abuse as
feckless, in that guilty plea rates are consistently
high, and prosecutors obtain convictions in a large
majority of the cases they file charges on.
• Neither should we view it as reckless, because
charges are filed, on average, less often than for other
felonies, and trials are relatively few.
• The variation in charges filed suggests district
attorneys’ offices differ based on case popula-
tion, programs, policy, philosophy, skills, or other
factors.
• Case flow of child abuse prosecution should be seen
as similar in many ways to prosecution of felonies in
general, because diversion, guilty plea, trial, and
conviction rates for child abuse cases were all within
6% of the rates for all felonies.
• Cases without charges filed and cases that end in
guilty pleas are more than 8 times more frequent
than trials and should receive attention more com-
mensurate with their numbers.
• Charging and conviction rates should not be used as
the sole measures of prosecution success because
they are subject to differences in case population,
and they may be unfair to good programs taking on
difficult cases.
• Additional measures of prosecution success need to
be developed.
• Diversion programs deserve additional examina-
tion as they are rarely used or discussed in the pro-
fessional literature.
• Standards for incarceration of defendants convicted
of child abuse crimes need further examination be-
cause the rate of incarceration was lower than for ev-
ery comparison category.
• We need to study existing rate data more thoroughly,
obtain more rate information from prosecutor data-
bases, and do research that goes beyond rates to ex-
amine a range of prosecution processes and
outcomes.
REFERENCES
†Boland, B., & Brady, E. (1985). The prosecution of felony
arrests, 1980 (NCJ-97684). Washington, DC: Department
of Justice, Bureau of Justice Statistics (BJS).
†Boland, B., Brady, E., Tyson, H., & Bassler, J. (1983). The
prosecution of felony arrests, 1979 (NCJ-86482). Washing-
ton, DC: Department of Justice, BJS.
†Boland, B., Conly, C. H., Mahanna, P., Warner, L., & Sones,
R. (1990). The prosecution of felony arrests, 1987 (NCJ-
124140). Washington, DC: Department of Justice, BJS.
†Boland, B., Conly, C. H., Warner, L., Sones, R., & Martin,
W. (1989). The prosecution of felony arrests, 1986 (NCJ-
113248). Washington, DC: Department of Justice, BJS.
†Boland, B., Logan, W., Sones, R., & Martin, W. (1988). The
prosecution of felony arrests, 1982 (NCJ-106990). Washing
-
ton, DC: Department of Justice, BJS.
†Boland, B., Mahanna, P., & Sones, R. (1992). The prosecu-
tion of felony arrests, 1988 (NCJ-130914). Washington,
DC: Department of Justice, BJS.
†Boland, B., & Sones, R. (1986). The prosecution of felony
arrests, 1981 (NCJ-101380). Washington, DC: Depart-
ment of Justice, BJS.
*Bradshaw, T. L., & Marks, A. E. (1990). Beyond a reason-
able doubt: Factors that influence the legal disposition
of child sexual abuse cases. Crime & Delinquency, 36(2),
276-285.
*Brewer, K. D., Rowe, D. M., & Brewer, D. D. (1997). Factors
related to prosecution of child sexual abuse cases. Jour
-
nal of Child Sexual Abuse, 6(1), 91-111.
†Brossi, K. B. (1979). Across-city comparison of felony case pro
-
cessing (LEAA Contract No. J-LEAA-016-76). Washing
-
ton, DC: Department of Justice.
at SAGE Publications - Full-Text Collections on February 9, 2009 http://tva.sagepub.comDownloaded from

Cashmore, J., & Horsky, M. (1988). The prosecution of child
sexual assault. Australian and New Zealand Journal of
Criminology, 21, 241-252.
Champion, D. J. (1988). Child sexual abusers and sentenc
-
ing severity. Federal Probation, 52, 53-57.
*Chapman, J. R., & Smith, B. (1987). Response of social ser-
vice and criminal justice agencies to child sexual abuse
complaints. Response, 10(3), 7-13.
*Cheit, R. E., & Goldschmidt, E. (1997). Child molesters in
the criminal justice system: Acomprehensive case-flow
analysis of the Rhode Island docket (1985-1993). New
England Journal on Criminal and Civil Confinement, 23(2),
267-301.
*Conte, J. R., & Berliner, L. (1981). Prosecution of the
offender in cases of sexual assault against children. Vic-
timology: An International Journal, 6(1-4), 102-109.
Cooper, H., & Hedges, L. V. (Eds.). (1994). The handbook of
research synthesis. New York: Russell Sage Foundation.
Cross, T. P., DeVos, E. D., & Whitcomb, D. (1994). Prosecu-
tion of child sexual abuse: Which cases are accepted?
Child Abuse and Neglect, 18(8), 663-677.
*Cross, T. P., Whitcomb, D., & DeVos, E. D. (1995). Criminal
justice outcomes of prosecution of child sexual abuse: A
case flow analysis. Child Abuse and Neglect, 19(12), 1431-
1442.
Cross, T. P., Martell, D., McDonald, E., & Ahl, M. (1999).
The criminal justice system and child placement in
child sexual abuse cases. Child Maltreatment, 4, 32-44.
*Cullen, B. J., Smith, P. H., Funk, J. B., & Haaf, R. A. (2000). A
matched cohort comparison of a criminal justice sys-
tem’s response to child sexual abuse: Aprofile of perpe-
trators. Child Abuse and Neglect, 24(4), 569-577.
Davidson, H. A. (1997). Courts and child maltreatment. In
M. E. Helfer, R. S. Kempe, & R. D. Krugman (Eds.), The
battered child (5th ed., pp. 482-499). Chicago: University
of Chicago Press.
*Davis, N. S., & Wells, S. J. (1996). Justice system processing of
child abuse and neglect cases. Final report. Washington,
DC: American Bar Association and National Institute of
Justice.
*DeFrancis, V. (1969). Protecting the child victim of sex crimes
committed by adults. Englewood, CO: The American
Humane Association, Children’s Division.
*DeJong, A. R., & Rose, M. (1991). Legal proof of child sex-
ual abuse in the absence of physical evidence. Pediatrics,
88(3), 506-511.
*Dolan, K. J. (1985). Evaluation of outcomes for reported
cases of child sexual abuse in Texas. Victimology: An
International Journal, 9(3-4), 383-397.
Dziech, B. W., & Schudson, C. B. (1989). On trial: America’s
courts and their treatment of sexually abused children.
Boston: Beacon.
*Faller, K. C., & Henry, J. (2000). Child sexual abuse: A case
study in community collaboration. Child Abuse and
Neglect, 24, 1215-1225.
Fridell, L. A. (1991). Intrafamilial child sexual abuse treat
-
ment: Prosecution following expulsion. Child Abuse and
Neglect, 15, 587-592.
*Goodman, G. S., Taub, E. P., Jones, D. P. H., England, P.,
Port, L. K., Rudy, L., et al. (1992). Testifying in criminal
court: Emotional effects on child sexual assault victims
(Serial No. 229). Monographs of the Society for Research in
Child Development, 57(5).
*Gray, E. (1993). Unequal justice: The prosecution of child sex-
ual abuse. New York: Free Press.
Harshbarger, S. (1987, March). Prosecution is an appropri-
ate response in child sexual abuse cases. Journal of Inter-
personal Violence, 2, 108-112.
†Hart, T. C., & Reaves, B. A. (1999). Felony defendants in large
urban counties, 1996 (NCJ-176981). Washington, DC:
Department of Justice, BJS.
Levesque, R. J. R. (1995). Prosecuting sex crimes against
children: Time for “outrageous” proposals? Law & Psy-
chology Review, 19, 59-91.
Light, R., & Pillemer, D. (1984). Summing up: The science of
reviewing research. Cambridge, MA: Harvard University
Press
Lipsey, M. W., & Wilson, D. B. (2001). Practical meta-
analysis. Thousand Oaks, CA: Sage.
*MacMurray, B. K. (1989). Criminal determination for
child sexual abuse: Prosecutor case-screening judg-
ments. Journal of Interpersonal Violence, 4(2), 233-244.
Martell, D. (2002). Child placement and criminal prosecu-
tion: A study of the relationship between criminal jus-
tice and protective service interventions in cases of
child abuse. Dissertation Abstracts International, 62(12),
4334A. (Publication No. AAT 3035629)
*Martone, M., Jaudes, P. K., & Cavins, M. K. (1996). Crimi-
nal prosecution of child sexual abuse cases. Child Abuse
and Neglect, 20(5), 457-464.
McEwen, T. (1995). National Assessment Program: 1994 sur-
vey results (NCJ-150856). Washington, DC: National
Institute of Justice.
McGregor, R. P. (1990). The prosecution of child molesters.
In R. L. Goldman & R. M. Gargiulo (Eds.), Children at
risk: An interdisciplinary approach to child abuse and
neglect. Austin, TX: Pro-Ed.
*Mennerich, A., Martell, D., Cross, T. P., & White, A. (2002).
Case and system predictors of prosecution of child abuse.
Manuscript submitted for publication.
Myers, J. E. B. (1994). Adjudication of child sexual abuse
cases. The Future of Children, 4(2), 84-102.
Nardulli, P. F., Eisenstein, J., & Flemming, R. B. (1988). The
tenor of justice. Chicago: University of Illinois Press.
National Children’s Alliance. (2001). [Number of NCA
member agencies]. Unpublished raw data.
Newberger, E. H. (1987, March). Prosecution: A problem-
atic approach to child abuse. Journal of Interpersonal Vio-
lence, 2, 112-117.
Ostrom, B. J., & Hanson, R. A. (2000). Efficiency, timeliness,
and quality: A new perspective from nine state trial courts
(NCJ-181942). Washington, DC: National Institute of
Justice.
Peters, J. M., Dinsmore, J., & Toth, P. (1989). Why prosecute
child abuse? South Dakota Law Review, 34, 649-659.
Cross et al. / PROSECUTION OF CHILD ABUSE 339
at SAGE Publications - Full-Text Collections on February 9, 2009 http://tva.sagepub.comDownloaded from

†Reaves, B. A. (1998). Felony defendants in large urban coun
-
ties, 1994 (NCJ-164616). Washington, DC: Department
of Justice, BJS.
†Reaves, B. A. (2001). Felony defendants in large urban coun
-
ties, 1998 (NCJ-187232). Washington, DC: Department
of Justice, BJS.
†Reaves, B. A., & Smith, P. Z. (1995). Felony defendants in
large urban counties 1992 (NCJ-148826). Washington,
DC: Department of Justice, BJS.
*Rogers, C. M. (1982). Child sexual abuse and the courts:
Preliminary findings. Social Work and Child Sexual
Abuse, 1(1), 145-153.
Sheppard, D. I., & Zangrillo, P. A. (1996, Winter). Coordi-
nating investigations of child abuse. Public Welfare,
54(1), 21-31.
*Smith, B. E., & Goretsky-Elstein, S. (1993). The prosecution
of child sexual and physical abuse cases. Washington, DC:
American Bar Association.
Stevens, G., & Fischer, D. G. (1992). Review and monitoring of
child sexual abuse cases in selected sites in Saskatchewan:
Studies on the sexual abuse of children in Canada. Ottawa:
Canada Department of Justice.
*Stroud, D. D.. Martens S., & Barker, J. (2000). Criminal
investigation of child sexual abuse: A comparison of
cases referred to the prosecutor to those not referred.
Child Abuse and Neglect, 24, 689-700.
*Tjaden, P. G., & Thoennes, N. (1992). Predictors of legal
intervention in child maltreatment cases. Child Abuse
and Neglect, 16, 807-821.
Trocme, N., MacLaurin, B., Fallon, B., Daciuk, J.,
Billingsley, D., Tourigny, M., et al. (2001). Canadian inci-
dence study of reported child abuse and neglect: Final report.
Ottawa, Canada: Minister of Public Works and Govern-
ment Services Canada.
Van Dam, C. V., Halliday, L., & Bates, C. (1985). The occur-
rence of sexual abuse in a small community. Canadian
Journal of Community Mental Health, 4(1), 105-111.
Whitcomb, D. (1992). When the victim is a child (2nd ed.)
(Contract No. OJP-86-C-009). Washington, DC: U.S.
Department of Justice, Office of Justice Programs,
National Institute of Justice.
Whitcomb, D., & Hardin, M. (1996). Coordinating criminal
and juvenile court proceedings in child maltreatment cases
(NCJ-161835 Grant No.92-IJ-CX-K034). Washington,
DC: National Institute of Justice.
Wilber, N. R. (1987). Dilemmas of addressing child sexual
abuse: The implementation of Chapter 288 in
Middlesex County. Dissertation Abstracts International,
48(7), 1889A. (Publication No. AAT 8722710)
Wright, L. (1995). Remembering Satan. New York: Vintage.
Referencesmarkedwithanasterisk(*)indicatestudiesincludedin
the meta-analysis.References marked with a dagger (†) indicate
studies included in the comparison.
Theodore P. Cross, Ph.D., is a research pro
-
fessor and director of the National Evaluation
of Children’s Advocacy Centers at the Crimes
Against Children Research Center at the Uni-
versity of New Hampshire. He has been con-
ducting research on system response to trou-
bled children for 17 years, including studies of prosecution
of child abuse, outcomes of foster care, and organization of
children’s services. He teaches advanced statistics at
Brandeis University and maintains a practice in clinical
psychology.
Wendy A. Walsh, Ph.D., is a research asso-
ciate at the Crimes Against Children
Research Center at the University of New
Hampshire for the National Evaluation of
Children’s Advocacy Centers. Her research
includes parents’ use of corporal punishment
and program evaluations of child protective services, child
welfare risk assessment systems, and family resource
centers.
Monique Simone, MSW, is a research
assistant at the Crimes Against Children
Research Center at the University of New
Hampshire for the National Evaluation of
Children’s Advocacy Centers. She is conduct-
ing research of interdisciplinary response to
nonoffending caregivers and their effects on investigations
in child sexual abuse cases and state criminal judiciary
response to sexual assault.
Lisa M. Jones, Ph.D., is a research assis-
tant professor of psychology at the University
of New Hampshire’s Crimes Against
Children Research Center and assistant direc-
tor of the National Evaluation of Children’s
Advocacy Centers. Her research has focused
on studying child abuse trends and evaluating community
response to child victims. She received her degree in clini-
cal psychology from the University of Rhode Island in
1999.
340 TRAUMA, VIOLENCE, & ABUSE / October 2003
at SAGE Publications - Full-Text Collections on February 9, 2009 http://tva.sagepub.comDownloaded from