Abstract
This study of attrition compares the prosecution of child sexual offenses reported while the complainant was still a child with those in which the report was delayed into adulthood; it also compares matters involving adult and young (under 18 years) suspects/defendants. It is based on an analysis of police and court administrative data in New South Wales, Australia over a 14-year period (2003–2016). Only one in five (21.6%) proceeded beyond the investigation stage. Criminal proceedings were more likely to commence when the alleged victim was 7–12 years old at the time of the incident, when the suspect was an adult and at least 10 years older than the victim, and also when the report to police was made when the victim was an adult. Just over half (55.5%) of the matters finalized in court resulted in a conviction. Cases in the higher courts were less likely to be dismissed and more likely to feature guilty pleas and convictions at trial than cases in the lower courts. The overall estimate is that only 12% of offenses reported to police resulted in a conviction, at a relatively stable rate over 14 years. These findings are consistent with those of comparable studies.
Child sexual abuse is widely recognized by police, prosecutors, and legal commentators as one of the most difficult crimes to investigate and prosecute. It is a crime that is almost always committed in secrecy and without witnesses. The main evidence, and in many cases the only evidence, is the child’s account of the offense. Children may be unwilling to disclose the abuse or have difficulty recounting the events (London, Bruck, Ceci, & Schuman, 2005). Delays in reporting are common, and many victims do not report it until well into their adult years (Parkinson, Oates, & Jayakody, 2010; Priebe & Sevedin, 2008).
There have been consistent and strong concerns across jurisdictions about the attrition of cases as they move through the system (Cross, Walsh, Simone, & Jones, 2003; Daly & Bouhours, 2010). Research over several decades in Australia, Canada, England and Wales, Scotland, and the United States indicates that when complaints of child sexual abuse are reported to the police, only a small proportion result in prosecution and conviction, with a substantial drop-off at each stage of the criminal justice process (Bunting, 2008, 2014; Daly & Bouhours, 2010; Eastwood, Kift, & Grace, 2006; Fitzgerald, 2006). The attrition estimates vary by jurisdiction, the data source (police, hospital referrals, prosecuting agency), and by the “starting point” (at report to the police or referral for prosecution). A range of characteristics of the child, alleged offense, and suspect (Cross et al., 2003; Daly & Bouhors, 2010) also contribute to the variation in reported attrition rates, as does the time period within which the data were collected. The current study explores the factors that predict attrition at different stages of the prosecution process using data over a 14-year period in one Australian state.
Points of Attrition
Decisions as to whether allegations of child sexual offenses proceed are made at various points in the criminal justice system—by police, prosecutors, and by the complainant. The first is the decision by the complainant or the child’s family to report the crime—or not—to the police or statutory child protection authority (Daly & Bouhours, 2010; Kelly, Lovett, & Regan, 2005). The proportion of offenses that are reported to the police or other authorities, and recorded, is unknown but estimated to be very low (Kelly & Karsna, 2017). For example, a Norwegian survey of more than 4,400 adolescent school students about sexuality and sexual abuse indicates that 65% of the girls and 23% of the boys self-reported some form of sexual abuse experience that included peer-related unwanted sexual contact or noncontact abuse; only 8.3% of these students had talked to a professional, and only 6.8% said the incident had been reported to the authorities or police (Priebe & Svedin, 2008).
Following a report being made to the police or to a statutory child protection agency, the next point of attrition is during the investigation, conducted by the police or by joint police–child protection teams. Police may decide for various reasons not to proceed, child complainants may not make a clear disclosure, or they or their parents may become unwilling to proceed (Bailey, Powell, & Brubacher, 2017; Bunting, 2008, 2014; Cross, De Vos, & Whitcomb, 1994; Daly & Bouhours, 2010; Fitzgerald, 2006; Kelly, Lovett, & Regan, 2005; Parkinson, Shrimpton, Swanston, O’Toole, & Oates, 2002). A number of studies in Australia, England, Ireland and a cross-national meta-analysis have quite consistently reported that about one in four to one in three cases are “lost at the police stage” with the complainant or the family “declining to complete the initial investigative process” (Bunting, 2008; Cross et al., 2003; Daly & Bouhours, 2010; Kelly et al., 2005). These and other studies indicate that a number of factors are associated with complainant willingness to proceed, particularly the relationship of the suspect to the complainant and the age of the child and the suspect (Bunting, 2008; Leach, Powell, Sharman, & Anglim, 2017; Pipe, Lamb, Orbach, & Cederborg, 2007). Preschool and young children have been found to be consistently less likely to disclose abuse in forensic interviews than older children (Bunting 2008; Hershkowitz, Lamb, & Katz, 2014; Lippert, Cross, Jones, & Walsh, 2009). Several studies have also reported a curvilinear association between the child’s age and the likelihood of the child disclosing and being willing to proceed, with cases involving children in mid-childhood (aged 7–11 years) more likely to proceed than those involving younger children and adolescents (Bunting, 2008; Leach et al., 2017; Walsh, Jones, Cross, & Lippert, 2010). Disclosure and willingness to proceed were also more likely in cases involving suspects unrelated to the child than immediate family members. When the suspect was similar in age (Leach et al., 2017) and the allegations involved similar-age boyfriends or intimate partners (Bunting, 2008), adolescents were least likely to be willing to proceed.
Even when the complainant and their family are willing to proceed, there are several reasons why the police may not do so. Police may be unable to identify or locate the suspect or decide that there is insufficient evidence to justify a prosecution. Less commonly, the police may conclude that the allegation is unfounded; for example, they reached that conclusion in 8.5% of cases involving family members compared with 4.8% of matters overall in a South Australian study (Wundersitz, 2003). Christensen, Sharman, and Powell (2016) found that the police did not lay charges in just over half of the 659 cases (51.1%) in which a child sexual offense was reported to them. The main reason was insufficient evidence (43.6%), followed by the child not disclosing the abuse in the early assessment interview or not providing sufficient detail about the offense (20.5%). Less common was a suspect not being identified (11.6%) or no offense being identified (8.0%). An earlier Australian study of attrition in sexual assault cases in 2004 in New South Wales (NSW) found that cases were more likely to proceed if the complainant was female, older than 10 at the time of the offense, knew the alleged offender, and made the report to the police within 10 years of the offense; overall, criminal proceedings commenced in only 15% of child sexual offense incidents (Fitzgerald, 2006).
The next point at which cases may “drop out” of the criminal justice system relates to prosecutorial decision-making about what charges should be prosecuted or whether the case should be discontinued. The main test is whether there is a reasonable prospect of a conviction based on the strength of the evidence and whether it is in the public interest to prosecute (Shead, 2014). A substantial body of research indicates that prosecutorial decisions are influenced by legal and nonlegal factors including the complainant’s age at the time of the offense, at complaint, and at prosecution (Cross et al., 1994; Ernberg, Tidefors, & Landström, 2016; Fitzgerald, 2006; Stroud, Martens, & Barker, 2000); the level of family support for the complainant and the psychological vulnerability of the complainant (Cross et al., 1994; Duron, 2018); the child’s cultural background (Bailey et al., 2017); and the nature and circumstances of the alleged abuse, such as the type, duration, and seriousness of the alleged offenses (Brewer, Rowe, & Brewer, 1997). The quality of the interview with the complainant is also a consideration in determining the adequacy and reliability of the evidence (Burrows & Powell, 2014).
Attrition may also occur after the matter is listed in court. Even at this stage, charges may be withdrawn, complainants may decline to proceed, and no evidence may be tendered by the prosecution. If the case proceeds, the defendant must decide whether to plead guilty or proceed to trial (Cross et al., 2003). This is often the subject of plea bargaining, an established but contentious practice that provides the incentive of a reduced sentence for an early plea, in return for sparing the complainant the stress of giving evidence and saving the state the expense of a trial (Flynn & Freiberg, 2018; Golding, Lynch, Malik, & Foster-Gimbel, 2018). In child sexual offense matters, defendants were more likely to plead not guilty when the charges were more serious (e.g., aggravated sexual assault) and more likely to plead guilty early rather than late when the charges were less serious (Ringland & Snow, 2014).
The final decision point is the outcome—in particular whether the defendant is convicted—in cases that proceed to a trial or hearing. There is a substantial body of research highlighting the factors that increase the likelihood of a conviction. The most important is the quality of the evidence, and in child sexual offense matters, this relies heavily on the evidence of the child complainant and the child’s capacity to provide a reliable and credible account under sustained questioning and cross-examination (Cashmore, 2017; Klemfuss, Quas, & Lyon, 2014).
In summary, matters may “drop out” at a number of stages from the initial report to police—in the identification and location of the suspect, the decision by police to lay charges, and by prosecutors to take the matter to court, by the defendant to plead guilty or not, and finally the verdict by the fact finder if the matter goes to trial or a hearing.
Many studies focus on 1 year’s data or use relatively small samples based on court files, prosecutors or hospital or child advocacy sources (see Cross et al.’s, 2003 meta-analysis). The current study uses information from the “population” of child sexual offenses from large police and court databases over a 14-year period in the most populous state in Australia. It examines the number of reports to police and the likelihood that criminal proceedings commenced, associated with the age of the child at the time of the alleged offense and at report, and the age of the suspect and relationship of the suspect to the victim. It also examines the plea and conviction rates in the court data. Studies using large-scale data over a decade or longer are uncommon in this area; however, exploring the trends in numbers of reports to police and court finalizations as well as the overall attrition rate over an extended period is important to provide some context and an indication of the variability that more time-limited studies may be subject to. Also uncommon are studies that differentiate between child sexual offenses reported while the complainant was still a child and those delayed into adulthood (Connolly, Coburn, & Chong, 2017) as well as those comparing cases involving young suspects (under 18) and those who are adults.
Method
De-identified data for all sexual offenses against children reported to NSW police over a 14-year period from 2003 to 2016 and cases finalized in NSW courts were obtained from the NSW Bureau of Crime Statistics and Research (BOCSAR: https://www.bocsar.nsw.gov.au/). The Bureau is a state government agency responsible for managing and “cleaning” the administrative data from the police (Computerised Operational Policing System [COPS]) database and the courts (court database) in NSW. The research had ethics approval from the Human Research Ethics Committee at the University of Sydney and research approval from BOCSAR.
The COPS data included information about the incident, the victim, and the suspect for all entries where the victim was under 18 years at the time of the alleged offense by year. The analyses of the police data distinguish between incidents reported while the complainant was still a child and those where the complainant waited until adulthood to report the alleged offense. The unit of analysis in the police data was the “reported incident,” recorded in the year in which it was reported to the police, rather than the year the alleged offense was committed. An incident is defined as an action reported to, and recorded by, police that falls into a single offense category (for instance, indecent assault), occurs at one location, during one uninterrupted time period, and involves the same victim(s) and offender(s) (Fitzgerald, 2006). “Proceeding to court” means that a suspect has been identified, charges have been laid, and a court attendance notice filed by police for criminal proceedings to commence.
The court data set included all cases in which sexual offenses against a child were finalized in NSW courts. A court appearance is defined as finalized when all the charges are dealt with by being proven, not proven, or dismissed; nonproven court appearances include those where the defendant is found not guilty and where all charges are withdrawn. 1 The court data include information about the defendant and the offenses but no information about complainants, so it is not possible to differentiate alleged child sexual offenses reported in childhood from those reported in adulthood in the analyses of the court data.
Sexual offenses against children were categorized into four main types of offense: sexual assault involving sexual intercourse/penetration—without consent or as defined as unlawful because of the age of the complainant and/or the position of authority or trust of the alleged offender in relation to the child, indecent assault—contact sexual offense not involving sexual intercourse/penetration, acts of indecency—noncontact sexual offenses, and child pornography. These categories are in line with the definitions and categories of sexual offenses used by BOCSAR, and these generic categories are a broad indicator of the seriousness of the offense (penetrative sexual assault is the most serious with the most severe penalties). More serious offenses are dealt with in the higher courts, and matters involving alleged offenders under the age of 18 years are dealt with in the Children’s Court or by therapeutic diversionary avenues unless it is a serious offense. Children under the age of 10 years are not liable to criminal action in Australia.
The plea rate in each court (higher courts or Local Court) was calculated as the proportion of defendants with finalized matters who pleaded guilty to at least one charge and proceeded to sentence in that court. The conviction rate in each type of court was calculated as the proportion of defendants whose matters were finalized in that court who pleaded or were found guilty of at least one charge.
Analysis
The analyses of reported incidents in this article are based on the contact offenses of child sexual assault and indecent assault and differentiate between reports made while the complainant was still a child (under 18 years) and those in which the report was delayed until adulthood (aged 18 and older). Excluding the incidents with missing age or date of birth, 93.3% of incidents in NSW involved one victim/complainant, 96.6% involved one suspect, and 88.8% involved one complainant and one suspect. As data for victims and suspects were matched with data for incidents, it was decided that only cases in which there was one victim and one suspect would be included for some analyses, particularly logistic regression, to ensure an unequivocal match between the three sources of information.
The court data are based on charges in finalized court appearances for each calendar year. The original court data file contained a record for each offense by each person; the individual records were aggregated for each person separately within years, courts, and offenses. This means that each record in the aggregated court data file concerns a single defendant associated with the pleas and finalized outcomes for different charges for that person within a year. There was a median of two charges for each finalized appearance, with 1,548 defendants facing combined charges of sexual and indecent assault. Only four individuals appeared twice in the same year.
Regression analyses
Logistic regression (Stata 15; StataCorp, 2017) was used to model the association between the probability of legal action commencing and the following factors: type of offense (sexual assault or indecent assault), the child’s age at the time of the offense, the child’s gender, the interval between the offense and the report being made to police (delay in reporting), and the year the report was made (2003–2016), simultaneously taking account of the effects of these other factors. Similarly, for the court data, logistic regression was used to predict the probability of a conviction (by verdict, plea, or proven offense) using the following factors: the type of offense (sexual assault and indecent assault), the interval between the (earliest) date of the offense and the finalization date for that matter, year finalized, gender of the defendant, and the age of the defendant at finalization. Analyses were carried out separately for each court (Children’s, Local, and higher courts).
For regression analyses, each data set (except that for the Children’s Court, which was relatively small) was randomly divided in two so that all the observations for a given episode (police data) or person (court data) occurred in one half sample only; a model was fitted on the first sample and tested on the second. Main effects were retained in both models regardless of their statistical significance. Interaction effects that were of interest because of the findings of previous research were subject to a model reduction process. An interaction effect that was not statistically significant at the nominated α level (.001 for the police data, .005 for the court data) when tested with the first sample was dropped before the model was tested with the second sample. Interaction effects that were not statistically significant when tested with the second sample were dropped from the final models presented here. The conservative α value and the “hold-out” samples were used to control Type I error in the light of the large size of the data sets and the number of effects tested. The method described by Williams (2000), implemented in Stata’s cluster option, was used to take account of the possible nonindependence of outcomes for incidents which were grouped into “events” in the COPS records, and for individuals who appeared in court in the same year in different offense categories.
Results
Table 1 presents the characteristics of 63,008 child sexual assault and indecent assault incidents reported to police during the period 2003–2016. There was a 34.6% increase from 4,116 reported incidents in 2003 to 5,540 in 2016. Most victim–complainants were aged 10 or older, with a mean age of 10.3 years (SD = 4.19) at the time of the incident. Indigenous complainants (8.6%) were overrepresented relative to the proportion of Indigenous people in the NSW population (3.5%; Australian Bureau of Statistics, 2018). Most commonly, the suspect was known to the child but unrelated (35.0%), a parent (16.5%), or a family member or member of the child’s household (16.8%). Only 5.5% were strangers. More than 1 in 4 of the 40,182 suspects (27.9%) were under 18 years of age at the time of the incident.
Characteristics of Sexual Offense Incidents Against Children in NSW 2003–2016.
“Cases” Proceeding to Prosecution
The threshold for prosecution is the police being able to both identify and locate the suspect. If the suspect is recorded by police as “not identified,” this does not necessarily mean that the police—or the victims—do not know who the suspect is; it generally means that the victim was unwilling or unable to make a clear disclosure. As Figure 1A shows, a suspect was identified in 63.6% of child reports to the police (sexual assault, 59.9%; indecent assault, 68.5%). A higher proportion of incidents reported in adulthood (84.7%) involved an identified suspect (ranging from 95% to 80% from 2003 to 2016). There were substantial differences between child- and adult-reported incidents associated with the age of the complainant, the age of the suspect, and the relationship between the suspect and the complainant. When the complainant was an adolescent aged 14 and older at the time of the incident, the suspect was less likely to be “identified” in child reports (56.7%) than in adult reports (85.4%). Similarly, when the incident involved a parent or other family members or a boyfriend/girlfriend, the suspect was recorded as “not identified” 3–4 times more frequently in child reports than in adult reports (26.3% cf. 6.4% for parents, 30.1% cf 9.1% for other family members, and 41.2% cf 12.5% for boyfriend/girlfriend). Unsurprisingly, there was a low rate of suspects “identified by police” when the suspect was a stranger or not known to the complainant at the time of the report (26.3%).

(A) Incidents of child sexual assault and indecent assault reported in childhood and adulthood to NSW Police 2003–2016. (B) Persons in finalized court appearances for child sexual assault and indecent assault reported in NSW 2003–2001.
Legal Action Commenced
The police commenced criminal proceedings in 29% of child reports and in 43.7% of adult-reported incidents in which a suspect was identified (see Figure 1A). Overall, only one in six (16.6%) child reports of sexual assault incidents and one in five (20.8%) indecent assault incidents over the 14-year period resulted in criminal proceedings. The percentages for adult reports “proceeding” were about double the child report rate: 32.0% of sexual assault and 47.3% of indecent assault incidents. Conversely, less than half the proportion of incidents involving a suspect under 18 at the time of the incident (and similarly at report) proceeded to court (16.5%) compared with 39.2% of suspects over the age of 18.
Predicting the Likelihood of Legal Proceedings
The predicted probability of legal action commencing for sexual assault and indecent assault differed by year of report (2003–2016), the type of offense (sexual assault vs. indecent assault), the age of the complainant at the time of the alleged incident, and whether the suspect was under 18 at the time of the incident (Table 2). Complainant gender was not significant. The likelihood of proceeding was significantly lower for suspects younger than 18 at the time of the incident, χ2 (1 df) = 255.27, p < .00001, OR = .36, and significantly higher for children who were at least 6 years of age at the time of the incident than for younger children, χ2 (3 df) = 261.59, p < .00001, ORs varied from 2.6 to 3.1. The two-way interaction involving the type of offense and whether the report was made in childhood or adulthood was significant, χ2 (1 df) = 12.72, p = .0004. As indicated earlier, adult reports of indecent assaults were more likely to proceed than adult reports of sexual assault (47.3% cf 32.0%) and more likely to proceed than child reports of either offense (sexual assault, 16.6% and indecent assault, 20.8%). The two-way interaction between the relationship of the suspect to the child and extent of delay in reporting was significant, χ2 (35 df) = 103.95, p < .001, but there was no clear or easily described pattern.
Reduced Logistic Regression Model Predicting Likelihood of Criminal Proceedings.
aTo save space, the coefficients for indicator variables have been omitted and reference categories for dummy coding are shown in square brackets. The full results are available from the corresponding author on request.
*p ≤ .05. **p ≤ .01. ***p ≤ .001.
The probability of legal action also varied with the difference in age between the suspect and the complainant by the age of the complainant. This effect was not investigated in the logistic regression because of empty cells (e.g., children under 10 cannot be charged with a criminal offense; see Figure 2). Legal action was least likely (.06–.12) when the suspect was within 2 years of the complainant’s age and most likely when the suspect was at least 10 years older than the complainant. There was a curvilinear relationship, with legal action more likely when the complainant was 7–12 years old and the suspect was at least 5 years older than the complainant (see Figure 2). The highest probability of legal action (.52) was when the complainant was 10–12 years old and the suspect was at least 10 years older.

Proportion of sexual assault and indecent assault offenses in which legal action commenced against a suspect by age difference between suspect and complainant and age of child complainant. The age of criminal responsibility in Australia is 10 years.
Court Outcomes
A total of 10,086 persons were prosecuted in relation to 24,266 charges for at least one offense of child sexual or indecent assault in finalized court appearances in the Children’s Court, Local Court, or in the higher courts over the period 2003–2016. A “person” is defined as a person charged with a group of one or more charges finalized by the court in the same year. The average number of charges per person has increased in all courts over the 14-year period, from 1.44 in 2003 to 3.11 in 2016. There was a 58% increase in the number of finalizations across all three courts from 647 in 2003 to 1,026 in 2016.
The vast majority of defendants were male (n = 9,898; 98.1%). Most were nonindigenous (82.3%) though Indigenous defendants (13.4%) were overrepresented (cf. 3.5% of the NSW population; Australian Bureau of Statistics, 2018). Only 1,371 (13.6%) defendants appeared in the Children’s Court. The higher courts dealt with the largest number of defendants (5,681, 56.3%; Figure 1B), with child sexual assault the most common offense in that court (3,175, 68.1%); 3,034 defendants (30.1%) were dealt with in the Local Court, with indecent assault the most common in that court (74.8%). The mean age of defendants at the time of the alleged offense was 36.1 years and 45.1 years at finalization in the higher courts, and 39.1 years and 43.8 years, respectively, in the Local Court. In the Children’s Court, the mean age at the time of the alleged offense was 14.9 years and 16.3 years at finalization. Most defendants who were under 18 at the time of the alleged offenses (67%) were dealt with in the Children’s Court, but 29.4% were dealt with in the higher courts and 3.6% in the Local Courts for more serious offenses.
In the higher courts, one in six (937, 16.5%) defendants had all charges dismissed without a hearing or “otherwise disposed of” in the period 2003–2016; 2,561 (45.1%) persons proceeded to sentence without a trial following an (early) guilty plea (at committal in the Local Court). More than one in three defendants (38.1%) proceeded to trial, and of those, 49.2% were convicted of at least one charge, including 57 defendants who had at least one (late) guilty plea but were found not guilty at trial on other charges. The overall conviction rate from 2003 to 2016 (the proportion of defendants whose matters were finalized in the higher courts who pleaded or were found guilty) was 63.8% for child sexual assault and indecent assault offenses (ranging from 50.5% in 2003 to 67.4% in 2013).
The plea rate was lower in the Local Court (29.2%) and Children’s Court (40.7%) than in the higher courts (45.1%; Figure 1B). The conviction rate was also substantially lower in the Local Court (42.0%) and in the Children’s Court (51.3%) than in the higher courts (63.9%). A high proportion of defendants had all charges dismissed, withdrawn, or not proceeded with in the Local Court (40.2%) and Children’s Court (29.1%) compared with only 16.5% in the higher courts. In the Local Court, 30.0% of defendants had a defended hearing, and of those, 40.8% were found guilty of at least one charge; 57.9% were acquitted with all charges dismissed. In the Children’s Court, 28.0% of defendants had a defended hearing which resulted in a “proven” offense for 29.9%; all charges were dismissed at hearing for 64.1%, and 23 (5.9%) had another unspecified outcome.
Predicting the Likelihood of a Conviction
Logistic regression was used to predict the probability of a conviction (by verdict or plea) in each court, including the following factors: the type of offense—sexual assault and indecent assault, year finalized, gender of the defendant, age of the defendant at finalization, and the interval between the (earliest) date of the offense and the finalization date (a proxy for delayed reporting).The final reduced model was significant for all three courts. The delay (time from the offense to finalization at court) was a significant factor in both the Children’s Court, χ2 (3 df) = 16.23, p = .001, ORs varied from .32 to 1.37, and Local Court, χ2 (4 df) = 27.02, p < .00001, ORs varied from .39 to .83, but the patterns were somewhat different. There was a marked downward trend in the proportion of persons who were convicted in the Local Court as the interval to finalization increased, from .43 for the shortest gap (less than a year) to a low of .24 for a gap of more than 10 years. In the Children’s Courts, the probability of a conviction was fairly consistent for delays of up to 5 years (.43–.51) but dropped to .20 for the very small numbers of defendants (51, 3.2%) whose matters were dealt with after a delay of 5 or more years in that court. The delay main effect did not reach the .005 criterion for the higher courts (p = .015). Taking all other factors into account, the probability of conviction was greater in the higher courts than in the Local Court and Children’s Court, and at all levels of delay. The other significant effect in the Children’s Court was the age of the defendant at the time of the trial, with older defendants more likely to be convicted.
Overall Attrition
An estimate of the overall proportion of incidents (sexual assault and indecent assault combined) reported to police (2003–2016) that resulted in a conviction was calculated by multiplying the proportion of reports proceeding to court by the “overall” conviction rate each year; it varied between 7.9% and 13.9% (see Figure 3). Legal action commenced for just over one in five (21.6%) of the 63,008 incidents reported to police over the 14-year period (18.4% of 52,352 child reports and 37.0% of 10,656 adult reports), varying between 16.8% in 2010 and 26.4% in 2003 (Figure 1A). All charges were dismissed in about 25.4% of cases across all three courts (16.5% in the higher courts, 40.2% in the Local Courts, and 29.1% in the Children’s Court). The conviction rate was 55.5% for persons whose matters were finalized in one of the three courts (varying between 41.6% in 2004 and 60.1% in 2013; see Figure 1B). This means that overall about 12% of sexual and indecent assault reports resulted in a conviction across the 14-year period, with some variation by year as Figure 3 shows. This calculation is based upon a 55.5% conviction rate multiplied by a 21.6% “proceed” rate (= 11.98%).

Number of incidents reported to police, number of finalized court appearances and plea, conviction and estimated “overall conviction” rates by year.
Discussion
The overall proportion of child sexual offenses reported to police that resulted in a conviction (around 12% of matters) is consistent with the findings of other studies reported earlier (Bunting, 2008, 2014; Daly & Bouhours, 2010; Eastwood et al., 2006; Fitzgerald, 2006; Wundersitz, 2003). The finding that only about one in five reports proceeded to court and that about half of these resulting in a conviction is also consistent with the findings from several studies that more cases “drop out” at the earlier stages, “before referral to prosecutors” than after (Daly & Bouhours, 2010; Kelly et al., 2005).
From Police Report to Prosecution
The first point of attrition involves the police identifying and locating a suspect in response to the information obtained from the complainant and during the investigation. Suspects were less likely to be “identified” in child reports than in adult reports, particularly when the complainant was an adolescent aged 14 or older at the time of the incident, and the suspect was either a parent, another family member, or a boyfriend/girlfriend. This probably reflects the child’s reluctance to make a clear disclosure and provide evidence against suspects when there is pressure from family members and lack of support from parents (London et al., 2005; Priebe & Svedin, 2008). This is consistent with the findings of a number of other studies in relation to the child’s age and relationship to the suspect (Bunting, 2008; Hershkowitz et al., 2014; Lippert et al., 2009) and also the importance of the support of the child’s caregiver and family (Duron, 2018).
Once the suspect is “identified” and located, there was little difference between child and adult reports in the likelihood that police would lay charges and start legal action against the suspect (just under one in three for both child and adult reports). In contrast with other studies, legal action was just as likely for male and female complainants (Edelson & Joa, 2010; Stroud et al., 2000).
The Age of the Complainant and the Suspect
Both the young age of the complainant at the time of the report and the youth of the suspect at the time of the incident were significant factors. Consistent with the findings of other studies, legal proceedings were least likely to be initiated when the child was aged 5 or younger (Bunting, 2014). For younger children, there appears to be a threshold (around 5–7 years) below which child victims are likely to be seen as “too young” for police or prosecutors to proceed or their families may be less willing for them to do so (Bunting, 2008; Daly & Bouhours, 2010). This is not surprising, given the likely evidentiary issues for young children (Cashmore, 2017; Shead, 2014).
The difference in age between the complainant and the suspect was also a significant factor, in combination with the complainant’s age. In general, the greater the age difference between them, the more likely the case was to proceed to prosecution. Legal action was more likely than not (52%), however, only for complainants aged 10–12 years, when the suspect was at least 10 years older (at least 22 years of age). This is consistent with the findings of several other studies concerning the likelihood of adolescents disclosing and of younger children providing sufficient evidence (Bunting, 2008; Leach et al., 2017; London et al., 2005; Walsh et al., 2010).
The age of the suspect at the time of the incident was also an important factor in whether the matter proceeded to prosecution. Incidents involving “persons of interest” under 18 were much less likely to result in legal action than for older suspects. The police and child protection response to a child or young “person of interest” is generally quite different, and appropriately so, depending on the facts, when it involves a peer-to-peer incident with a small age difference than when the suspect is an adult (Royal Commission into Institutional Responses to Child Sexual Abuse, 2017). Further, no police action is possible when the suspect is a child below the age of criminal responsibility (10 years of age in NSW). With the option of other diversionary processes and appropriate exercise of discretion, child-against-child reports are the least likely to result in legal action, especially when there is little age difference. This is consistent with Ferrante, Clare, Randall, and Boyd’s (2017) finding using similar data in NSW for the period 2010–2014 that adult-to-child offenses were 3 times more likely (33%) to proceed to court than child-to-child incidents (11%). In that study, abuse by another child, at least 3 years older than the complainant, was more likely to proceed than adolescent peer-to-peer incidents with less than a 2-year difference in age, and child-to-child incidents with a 1-year difference in age. That study too found a similar and unexpectedly high proportion of reports involving children and young people as suspects (32.1% for the period 2010–2014 compared with 28% for the period 2003–2016).
Delayed Reporting
Adult reports of both sexual and indecent assault were more likely to proceed than child reports. This is contrary to the expectation that delayed reporting and historical matters reported years after the incidents would result in low rates of legal action because of the evidentiary problems with degraded evidence or unavailable witnesses (Connolly et al., 2017; Hamer, 2010; Shead, 2014). The higher rate for adult reports proceeding is, however, largely a function of suspects being more likely to be “identified” in adult reports. It is likely that adults coming forward to report offenses that occurred in childhood may be more willing and determined complainants. Adult reporters are also likely to have less daily exposure to the potential defendant and a non-offending caregiver and therefore under less pressure to recant. Mandatory reporting laws also mean that contemporaneous child sexual offenses are likely to be reported whether or not the child is a willing or capable witness, whereas those who delay their report until they are adults are more likely to be making a considered choice. Adult reporters are therefore likely to constitute a selected group of the unknown number of cases of child sexual offenses that occurred some years before.
From Prosecution to Conviction
Once the police have laid charges, the responsibility for the carriage of a criminal prosecution is in the hands of a prosecutor, who has to make a call on whether there is a reasonable chance of a conviction and that prosecution is in the public interest. At this stage, the prosecutor can change the charges, and enter into plea and charge negotiation with the defendant’s lawyer, or decide not to proceed at all. In the higher courts, cases were less likely to be dismissed and more likely to feature guilty pleas and convictions at trial, compared with cases in the lower courts. This probably reflects the substantially higher penalties available in the higher courts that provide a greater incentive to negotiate on pleas and charge. It may also reflect the scrutiny and filtering that these cases receive through the investigation and committal process as well as the considerable efforts and special measures to support child witnesses and reduce the stress of giving evidence in the higher courts (Cashmore, 2017).The relatively high rate of cases being dismissed in the Local Court and the low plea rate in this court compared with the higher courts may indicate the need for more attention to the preparation of the matter, to plea negotiation, and support for the complainant in the lower courts, albeit the less serious charges being dealt with.
Overall, the plea and conviction rates in this study are consistent with the figures reported by Daly and Bouhours (2010, pp. 583–586) and Fitzgerald (2006). As Daly and Bouhours (2010) and Cross, Walsh, Simone, and Jones (2003) pointed out, it is important to be aware that different studies often use different measures and use different denominators in calculating charging and conviction rates. The high rates of plea and conviction that Cross et al. cite—an average of 82% for plea rates and 94% for conviction rates—excluded cases that were not carried forward (not dismissed or withdrawn). Using the same approach as Cross et al., the equivalent plea rates in the current study would be 61.8% in the higher courts, and 57.9% and 57.5% in the Local and Children’s Courts, still considerably lower than Cross et al.’s figure of 82%. Similarly, the comparable conviction rate in the higher court would be 80.2%, and 76.6% and 72.7% in the Local and Children’s Courts, again considerably lower than Cross et al.’s 94% figure. The 24 studies included in Cross et al.’s meta-analysis were conducted from 1969 to 2002, mostly in the 1980s and the 1990s, and largely in the United States. The most recent figures in the current study are 20–30 years on, in Australia. They reflect an increasing number of finalized appearances and it is possible that, again as Cross et al. suggest, a more “zealous” commitment to prosecution. Following legal reforms over the past several decades and the impact of the Royal Commission into Institutional Response to Child Sexual Abuse in Australia, this may not translate into higher plea and conviction rates as police and prosecutors pursue matters with younger children and with weaker evidence. This was evident in the 1990s in NSW when legislative reforms made it easier to prosecute cases in which the child was too young to be a “sworn” witness: The number of cases increased 4-fold from 1982 to 1992, but both the plea and conviction rates dropped significantly (Cashmore, 1995).
Limitations
The nature of the police and court data do not allow us to track matters from reporting to police, through the investigation and prosecution process to court, and to finalization at court via conviction and sentencing (Fitzgerald, 2006). Further, as pointed out earlier, the court data do not include information on complainants such as their age at the time of the incident, at report to police, or at prosecution, so it is not possible to directly compare court matters where reports were made in childhood and adulthood or to use the child’s age to predict the likelihood of a guilty plea or conviction.
While there is a fair degree of consistency with other findings on the attrition of cases as they work their way through the criminal justice system, these data are drawn from one Australian state within a certain time period.
These databases also do not provide information about how police and prosecutors decide which cases will proceed and what factors they take into consideration in their assessments of the quality of the evidence and the vulnerability of the child and capacity to handle the stressfulness of the process (Duron, 2018). Further research is required to understand how these decisions are made, over and above the characteristics of the complainant, suspect and the type of offense, commonly associated with a higher likelihood of criminal prosecution and conviction (Duron, 2018; Ernberg et al., 2016).
Conclusions
The way in which reports to police progress through the early stages of the criminal justice system varied with the age of the child—both at the time of the alleged offense and when the report is made to the police. It also varied with the relationship of the suspect to the child, the type of offense, the age of the suspect, the difference in age between the alleged offender and the child, and by the year the report was made. In the court process, the conviction rate varied significantly across courts, with the time to finalization (a proxy for delayed reporting of the offense), and with the age of the defendant. The main “dropout” occurred in relation to the decision by police as to whether charges were laid, with criminal proceedings commencing in only about one in five incidents reported to the police. Once matters were referred to prosecutors and reached the courts, there was still a substantial dropout, depending on the court (16.5–40%). About 40–64% of defendants were convicted, with the highest rate in the higher courts dealing with the more serious offense of sexual assault. The plea and conviction rates were substantially lower than the 80–90% rates reported several decades ago in NSW (Cashmore, 2017), when only a small number and proportion of cases passed the competency and corroboration hurdles.
The dropout rate or attrition should not necessarily be seen as a failure of the criminal justice system. The function of the police, prosecutors, and the courts is to assess the evidence and to some extent the impact of the process on victims and their capacity to withstand what can be a grueling process. Not all reports will involve criminal offenses that should proceed to prosecution. Some that are the subject of mandatory reports may not warrant criminal action and some may be more effectively dealt with by child protection agencies and diversion processes, particularly those involving youth under 18. The system clearly fails, however, when poor interviewing and investigation by police fail to elicit relevant evidence and leave that child and others at risk. It also fails when stressful prosecution and court processes intimidate complainants, making children and families unwilling to be involved, and making it difficult or impossible for the victims of child sexual abuse—as child witnesses or as adults in delayed reporting matters—to provide reliable evidence. It is not clear, however, exactly how police and prosecutors decide which cases should proceed—and to what extent the cases that do not reach the courts are matters that would and should have resulted in an appropriate plea or conviction. Those are the attrition “failures.”
Supplemental Material
Supplemental Material, Full_regression_tables_-_supplementary_tables - Fourteen-Year Trends in the Criminal Justice Response to Child Sexual Abuse Reports in New South Wales
Supplemental Material, Full_regression_tables_-_supplementary_tables for Fourteen-Year Trends in the Criminal Justice Response to Child Sexual Abuse Reports in New South Wales by Judith Cashmore, Alan Taylor and Patrick Parkinson in Child Maltreatment
Footnotes
Authors’ Note
We appreciate the assistance and discussions with the NSW Bureau of Crime Statistics and Research, NSW Police and the NSW Office of the Director of Public Prosecutions, and the very constructive comments of the journal’s editor and the peer reviewers.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The research was funded by the Australian Research Council Discovery Grant DP 160103688, following earlier funded research for the Royal Commission into Institutional Responses to Child Sexual Abuse in Australia.
Supplemental Material
Supplemental material for this article is available online.
Note
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
