Using Multiple Informants to Assess Child Maltreatment: Concordance Between Case File and Youth Self-Report

Abstract

To understand the psychosocial implications of child maltreatment, methods used to document prevalence must be clear. Yet, rates of maltreatment found in child self-report are generally inconsistent with data found in case files from state social service agencies. Although self-reports and case file reports of abuse disagree on occurrence of specific events, it is unclear if reporters agree when overall categories of abuse are considered. This study investigated differences between case file and youth report of abuse by examining four types of abuse: physical, sexual, neglect, and psychological, in a within-subjects design using a sample of 97 youth in foster care aged 8 to 22. Case files were coded for the presence of any indication of each type of abuse. Self-report of abuse was also assessed for any indication of each type of abuse. Results indicated that, overall, youth reported more physical and psychological abuse, and younger youth reported more sexual abuse than documented in their file. Implications for research and service provision for maltreated youth are discussed.

Determining rates of maltreatment in youth is not easy and poses a challenge for those who seek to understand the scope of the problem. Yet, knowing a child’s maltreatment history is important. If the degree and nature of a child's abuse history remain largely unknown, service providers are unable to triage services so that children receive pertinent care in a timely manner. Regrettably, reported rates of maltreatment can differ depending on the methods used to gather the data and the sample from which data are drawn. Case file documentation of reports made to Child Protective Services (CPS) and youth/caregiver reports of maltreatment are two common sources for information and samples on which rates are determined. For example, the Fourth National Incidence Study of Abuse and Neglect (NIS-4; Sedlak et al., 2010), using data from reports made to CPS, found that 2% of children in the general population are maltreated each year. Other researchers in studies using youth and/or parent-report found that the rate of maltreatment in the general population is higher, around 13% (Finkelhor, Ormrod, Turner, & Hamby, 2005). Such discrepancies do little to explain the actual rate of maltreatment and make it difficult to determine the need for services but are not surprising when different methods are used. To move beyond documenting that different methods provide discrepant results, the present study sought to examine in what ways maltreatment information diverge when defined by two most common methods of maltreatment reports: CPS data and child-self-report.

The authors use the term “child maltreatment” to represent cases of child abuse and/or neglect, and use “abuse” and “child maltreatment” interchangeably. The definition provided by the Centers for Disease Control and Prevention for maltreatment was used: “Any act or series of acts of commission or omission by a parent or other caregiver that results in harm, potential harm, or threat of harm to a child” (Leeb, Paulozzi, Melanson, Simon, & Arias, 2008, p. 11).

Generally, researchers use one method or the other – CPS data or child self-report – to determine if and how children have experienced maltreatment. Because it is possible that the information one gets when using the child’s case file can be different compared to self-report, it is important for researchers and service providers to know not just that the two methods may give different answers, but in what ways the reports of maltreatment differ, especially if the goal is to compare findings. Each source of information may provide unique material since the requirements for an indication of child abuse may be dissimilar for case file report (i.e., preponderance of evidence) compared to child self-report (i.e., child’s memory for the event). Whereas most studies document or compare rates and types of abuse from one source or the other, the present study examined within child differences. The goal was to identify how what each child reports may be similar or not to what their case file indicates about their history of abuse – and thus how treatment recommendations could differ if relying on one source or another.

Whether the reporter is a state agency or the child victim, discrepancies in reported rates of abuse exist for various reasons. Reasons include the often covert nature of maltreatment, lack of reporting by mandated reporters (Fraser, Matthews, Walsh, Chen, & Dunne, 2010), definitional incongruence regarding what constitutes maltreatment between states (Leeb et al., 2008), the tendency of most CPS agencies to reduce the official determination of maltreatment (i.e., reason for removal) to one type of abuse, and that many self-report methods are not designed for children younger than 12 years old (Finkelhor et al., 2005). Miller-Perrin and Perrin (2013) have described the process that leads to the field’s understanding of the prevalence of child maltreatment as a “funnel” with the largest number of actual events of child maltreatment (largely unknown) at the top, followed by cases that come to the attention of mandated reporters, cases of maltreatment reported to CPS, cases that are “screened-in” by CPS workers and thus counted in official reports, and then finally substantiated cases of child maltreatment, representing the smallest number, at the bottom of the funnel. Unfortunately, what is not documented is not treated, perhaps handicapping the effectiveness of efforts to provide services.

Case Files as a Reporting Method

To determine the extent of a child’s maltreatment history, data from case file reports from social service agencies are commonly used (Feiring & Zielinski, 2011). Case files are a likely go-to source given that case files contain reports of alleged maltreatment made to CPS via “hotline calls” to the agency (DePanfilis, 2006), outcomes of these hotline calls, and conclusions of the reports (i.e., substantiation vs. unsubstantiation), amongst other information. However, there are drawbacks to relying on case files as the ultimate information source. First, specific information contained within case files can differ from county to county, and thus the frequency and severity of abuse may be difficult to compare across samples. Second, although most studies on child maltreatment include youth who have substantiated cases of abuse and/or neglect as the indicator or confirmation that their sample is indeed composed of maltreated youth (Barnes, Noll, Putnam, & Trickett, 2009; Shenk, Noll, Putnam, & Trickett, 2010), this approach may miss important information about a child’s abuse history, as some forms of abuse are easier substantiate than others (Knight et al., 2000).

For example, neglect and physical abuse are often more easily identified by an outside observer than sexual abuse, as the latter is often a secret with physical signs being potentially difficult to document. It is also possible that there is little evidence to support the unsubstantiated child abuse claims, and research is needed to determine if the reason claims are unsubstantiated are because of the type of abuse, reporter, or the rules of evidence for documenting abuse. Moreover, even substantiated reports of abuse could contain false positives (United States Department of Health and Human Services, 2010). NIS-4 data indicate that the difference between child protective agency findings for substantiation or unsubstantiation is often accounted for by variability within the CPS system (not in the child) – where there are differences in protocols used for assessments and investigations that vary from case worker to case worker. As a result, research on youth exposed to maltreatment based only on substantiated reports may represent a subset of children who experienced specific types of abuse (US DHHS, 2010).

The inclusion of both substantiated and unsubstantiated claims may provide a closer estimate of a child’s actual maltreatment experiences if case files are to be used as a reporting method. In fact, evidence suggests that children with unsubstantiated claims are behaviorally and developmentally similar to those children with documented histories of abuse and neglect (Hussey et al., 2005; Leiter, Myers, & Zingraff, 1994), and children with unsubstantiated reports of child maltreatment have a similar risk for future maltreatment as those with substantiated reports (Drake, Jonson-Reid, Way, & Chung, 2003; English, Marshall, Coghlan, Brummel, & Orme, 2002; Kohl, Jonson-Reid, & Drake, 2009). It is concerning that the outcomes of research that utilizes only substantiated claims of abuse may not reflect the child’s total abuse experiences and equally concerning that false positives and negatives likely remain even when both substantiated and unsubstantiated claims are used. Thus, some researchers have also examined child self-report in an effort to accurately capture maltreatment experiences.

Youth Self-Report as a Reporting Method

Asking youth directly to report on their history of abuse is another method, albeit less common, used by researchers to gather information on the rates of child maltreatment (e.g., Amaya-Jackson, Socolar, Hunter, Runyan, & Colindres, 2000; Finkelhor et al., 2005). Perhaps surprisingly, youth are not necessarily the primary source of information about abuse experiences or rates of abuse in most maltreatment research. And – just like with case files – there are problems inherent to using children as a reporting method. For most types of abuse, although youth victims are the primary witnesses, they might not always be the best reporters. It would be easy for researchers if children could consistently report on the events in their past. Admittedly, the task of recounting one’s abuse history to researchers may not always be challenging for youth who have had to repeat their report to law enforcement and child welfare workers. Yet developmental level can hamper the accuracy and consistency of reports from youth, with younger children being less likely to accurately recall abuse and with retrospective reports often proving unreliable (Greenhoot, 2011). As a result, having a child recount their abuse experiences is not universally valued as a research method.

Researchers may be reluctant to ask child victims directly about previous abuse experiences due to a desire to not negatively impact the child’s mental health by asking them to recall a sensitive experience (Knight et al., 2000). Children – regardless of whether they have experienced maltreatment – are considered to be members of a vulnerable population when it comes to research. Research with children, and thus particularly children with maltreatment histories, should be conducted with increased caution and with attention to ethical guidelines. However, despite reasonable concerns, research has not indicated that recollecting traumatic events during research studies causes long-term distress (Legerski & Bunnell, 2010) – albeit data from child samples remain scarce.

Regardless of difficulties with using youth self-report of maltreatment, there is also some indication for valuing youth report as a research method. Even though precise recall of events may be a concern, recent research suggests that childhood traumatic experiences may be more easily and accurately remembered than non-traumatic, mundane events (Alexander et al., 2005; Cordon, Pipe, Sayfan, Melinder, & Goodman, 2004). Additionally, after a review of the literature on traumatic experiences and memory, Howe, Toth, and Cicchetti (2006) concluded that there is little evidence that maltreatment experiences have any negative impact on the memory processes of children. Evidence also suggests that children age 8 and older can consistently remember important personal health information, even for fairly benign events (Riley, 2004). Further, self-report of abuse could be beneficial in research because the self-report method may identify instances of maltreatment that have gone undetected by CPS, particularly those events that are more chronic in nature (Kendall-Tackett & Becker-Blease, 2004). Finally, it is also possible that youth may benefit from sharing their experiences. Treatment approaches such as Trauma-Focused Cognitive Behavioral Therapy (TF-CBT) include instruction on creating a trauma narrative, where the child is encouraged to specify the details of their abuse as an important part of the recovery process.

Multiple Informants and Maltreatment Research

Although there is a need for more research to address how information gathered from case file report and child self-report of child maltreatment might differ, some work exists. Overall, findings suggest that when youth are asked about their abuse, the number of events reported is significantly higher and generally discrepant with their official record or case file (Everson et al., 2008). One study found that rates of psychological abuse have the least amount of agreement between case file report and child self-report (6% of the sample was identified as psychologically abused by CPS report compared to 39% that was identified by self-report), followed by physical abuse (5% compared to 21%) and then sexual abuse (2% compared to 9%, Everson et al., 2008).

Data regarding discrepancies in reports of neglect are scarce. One study using prospective methods, which included a review of CPS reports as well as obtaining maltreatment reports from youth and their caregivers throughout their participation in an ongoing longitudinal study, indicated that prospective reports uncovered the most occurrences of neglect (22 cases) but that the best method of identification was to use both prospective and retrospective reports (which overall identified 30 cases of neglect; Shaffer, Huston, & Egeland, 2008). Although these findings are important, it is still unclear what discrepancies may be consistently found between case file and youth self-report. Clearly, youth self-report of maltreatment will be retrospective to some degree but arguably less so than adult self-report of maltreatment. Another limit to the current understanding of discrepancies between case file and youth self-report of abuse is that although certain that discrepancies exist, it is important to investigate the pattern across types, that is to determine if one source may be more likely to indicate one type of abuse (e.g., sexual abuse), whereas another is more likely to indicate another (e.g., neglect).

Because youth report and case file report of abuse are likely discrepant and because the discrepancy may be related to the type of abuse in question, the present study sought to detail the nature of the differences by providing an analysis of the types of abuse reported by youth and reported by the case file. The study sought to clarify the magnitude and direction of the difference between case file report and youth self-report of abuse for four types of abuse: physical, sexual, psychological, and neglect, to better understand the benefits or drawbacks to using just one of these sources to inform research findings or treatment decisions.

It was hypothesized that in general, children would report more abuse events than found in their case file across physical, sexual, and psychological abuse. Previous research has shown that physical and psychological abuse are reported more frequently by child self-report than case file report (Everson et al., 2008). Although reports of sexual abuse have not been previously found to be discrepant, it was also hypothesized that more sexual abuse would be reported via youth self-report due to the often covert nature of sexual abuse events. Further, even though the literature thus far does not provide support for a directional hypothesis regarding the discrepancy between reports for neglect, it was expected that children would report fewer neglect events than found in their case file. The reason for this hypothesis was the idea that children might be less aware that their caretaking has been neglectful than a mandated reporter who might, for example, easily recognize that a child’s living conditions are poor enough to be considered neglect. Age of the child was also explored for its relation to concordance between self-report and case file report. Age was considered because younger children may be unreliable reporters of life events and that older children, providing retrospective reports of abuse, may also be unreliable reporters due to the potential increased time between the abuse event and the report (Greenhoot, 2011).

Method

Participants

Participants were 97 youth in foster care in a mid-western state enrolled in the SPARK Project (Studying Pathways to Adjustment and Resilience in Kids). The SPARK Project is a research initiative designed to assess the behavioral and emotional outcomes of youth exposed to child maltreatment and placed in foster care. The youth were between the ages of 8 and 22, with a mean age of 13.51 years (SD = 3.18 years). Exclusionary criteria included a previous diagnosis of Mental Retardation or Autism due to the reliance on youth self-report during the data collection procedures. The sample consisted of 43 females and 54 males, and 54% of the sample self-identified as African American, 24% as Caucasian, 12% as multiracial, and 5% as Hispanic.

Measures

Case file report of abuse

The Modified Maltreatment Classification System (MMCS; English & Longitudinal Studies of Child Abuse and Neglect [LONGSCAN] Investigators, 1997) was used to code the types of abuse in each child’s CPS case file. The MMCS was designed as a way to classify and categorize information from child protective service records (English et al., 2002). The MMCS includes codes for abuse type (i.e., physical, sexual, psychological, and neglect), severity, and frequency of abuse (English & LONGSCAN Investigators, 1997).

Self-report of abuse

Child self-report of abuse was collected by adapting the MMCS items from a list to a self-report format (e.g., items like “the child was hit or kicked in the face” were changed to “have you ever been hit or kicked in the face?”). The self-report forms contained all of the abuse categories and subtypes from the MMCS and were asked in two parts: one question asked if the abuse ever occurred and, if so, the second question asked how frequently the abuse occurred. Only variables pertaining to whether or not abuse occurred within each category, not frequency of abuse, were used in analyses.

The Self-Report of Physical Maltreatment and Assault (PHYA) includes 18 questions about lifetime experience of physical abuse by any person in the child’s life (e.g., “In your lifetime, how often did someone kick or punch you?”). Self-report of neglect was measured by the About My Parents neglect scale (AMPA). The AMPA includes 25 questions about lifetime experience of neglect by the child’s caregivers, including their biological and/or foster parents (e.g., “In your lifetime, how often did your parents care if you did bad things, like shoplifting?”). Self-report of sexual abuse was determined using the Sexual Abuse and Assault Scale (SARA). The SARA includes 12 questions about lifetime experience of sexual abuse (e.g., “In your lifetime, how often has someone forced you to look at their sexual parts?”). Finally, the Psychological Maltreatment (PSMA) measure from the MMCS was used to measure psychological abuse. The PSMA contains 26 questions about psychological maltreatment by a caregiver (e.g., “In your lifetime, how often has anyone ever blamed you for their own problems?”). Although children also reported on the frequency of abuse, for the present study, only the presence or absence of any indication of each type of abuse was used in analyses (for more information for on measure development see Knight et al., 2000). Each self-report scale of abuse has a different number of questions, however, each question in the self-report scale corresponded to a specific type of abuse coded from the MMCS, one of the most commonly used measures of abuse types and forms (English & the LONGSCAN Investigators, 1997). One final note is that the abbreviations of the scales (e.g., PHYA, SARA) are not direct acronyms of the scale names (e.g., Self-Report of Physical Maltreatment and Assault, Sexual Abuse and Assault Scale) but are the abbreviations used by the scale authors, likely for consistency across the abbreviations used for the four scales.

Procedure

The SPARK project is a federally-funded, longitudinal study of resiliency and adaptation of youth in foster care ages 8 and older. Because research on youth in foster care requires several levels of permission, the SPARK project was approved by the state social service agency and the circuit court legally responsible for the welfare of youth in foster care, in addition to the institutional review board for the university and research review board for the state social service agency. The state CPS agency provided consent for the youth to participate and for the research team to access each child’s case file. Youth in the study were read an assent form by members of the research team. The state, court, and each participant was provided comprehensive information about the research protocol as well as information regarding confidentiality and the limits to confidentiality should any ongoing maltreatment or suicidality/homicidality be reported during the study. All research assistants who administered the surveys were graduate-level clinical child psychology students and were supervised by a licensed, board certified clinical child psychologist.

If any answers provided by participants indicated current harm, the computerized survey provided a “flagged items” page that would alert research assistants to follow up with participants and then make a report to CPS of current maltreatment and/or ensure the child’s safety. Out of all of the participants, approximately three indicated current abuse that was then reported to CPS. Because all children in the study were currently in the state’s custody, and because part of the study’s purpose was to obtain the most accurate information possible regarding the frequency of past history of abuse (with an emphasis on confidentiality for past reports of abuse to help ensure honest reporting), reports of past abuse were not included in follow-up. However, with all participants, regardless of endorsement of current harm, researchers utilized a three-part debriefing process in which the research assistant met with the caregiver alone, the youth participant alone, and then the youth and caregiver together to ensure all questions about the study were answered and concerns were addressed.

Each child was also assessed during data collection for their mood and well-being to ensure that any distress from answering questions was resolved. Further, a follow-up phone call was made within 48 hours to all participants to assess study-related distress (see Jackson, Gabrielli, Tunno, & Hambrick, 2012, for further information regarding the project’s methods). None of the children in the study reported any adverse effects from participating in the study during the debriefing or at the 48 hour follow-up. All caregivers were mailed information on mental health resources approximately one week after participating in case they felt the need for additional support following participation. As further evidence of the lack of harm from participating, the SPARK project retains over 90% of participants at each subsequent time point including youth and caregivers reported to CPS.

The SPARK project collects data from caregivers as well as youth and CPS records regarding the child’s exposure to abuse, emotional and behavioral health, academic functioning, and physical health over three time periods. For this study, only the information from the child self-report and CPS report of abuse from time one is included due to ongoing data collection.

To control for reading level, all questions were read aloud by a laptop computer over headphones via an audio computer-assisted self-interview program (ACASI). The ACASI system is also designed to promote comfort when answering questions about potentially sensitive experiences, making it more likely that participants will disclose information about their abuse than when asked similar questions in person (Kim, Dubowitz, Hudson-Martin, & Lane, 2008).

Each child’s case file was also collected as a part of the data collection procedures for SPARK. The information in the case file was coded using the MMCS (English & LONGSCAN Investigators, 1997) by a former employee of the state social service agency responsible for foster care who had no previous or current contact with any of the children in the study. The coder reviewed the case file and coded all reports of physical, sexual, and psychological abuse, as well as neglect for both unsubstantiated and substantiated claims of abuse. Following coding, four dichotomous variables were created to represent the presence or absence of any indication of the four abuse types in the file. Each child had a possible score of 1 (present) or 0 (absent) for each of the four abuse types. Similarly, if the child indicated a positive response to any of the abuse items in any type (e.g., physical, sexual) while completing the abuse-related measures on the ACASI, the child received a score of 1 for that type of abuse. If no abuse items within a given type were endorsed by the child, the child received a score of 0 for that abuse type.

Results

To test the study predictions that the rates of all four abuse types found in the case file and self-report would differ, frequencies were examined to determine the rates of abuse found per type of report. Results indicated that youth reported more exposure to abuse across all types than was reported in their case files. For physical abuse, 95% of the sample was identified as physically abused by self-report compared to 71% of the sample by case file report. For sexual abuse, 49% was identified by youth self-report compared to 41% by case file report. For neglect, 90% was identified by youth self-report whereas 85% was identified by case file report. Finally, for psychological abuse, 99% of youth were identified by youth report and 51% by case file report. Exact rates found per reporter are depicted in Figure 1. Next, to investigate whether there was a significant difference between child self-report and case file report and to understand how the reports differed (e.g., significant mismatch between case files and child self-report), a series of nonparametric, related samples McNemar tests were conducted. Given that less than 5% of the data were missing, missingness was managed with list-wise deletion (Graham, 2009).

Figure 1.

Child self-report and case file report by type of abuse

For physical abuse, the distribution of values found across case file report and child self-report significantly differed (χ² (1) = 17.46, p < .001). Specific results were that case file records indicated that 30 youth in the sample had no history of physical abuse; however, 29 (97%) of these 30 children self-reported physical abuse. Interestingly, of the 67 children whose case file indicated exposure to physical abuse, almost all (63; 94%) also self-reported a history of physical abuse (see Table 1), indicating that overall more children reported some kind of physical abuse experience than was reported in their case file.

Table 1.

Congruence Between Case File Report and Child Self-Report

Type of Abuse	Case file report	Self-report
Physical abuse		Yes	No
	Yes	63	4
	No	29	1
Sexual abuse		Yes	No
	Yes	18	20
	No	31	28
Neglect		Yes	No
	Yes	72	8
	No	15	2
Psychological abuse		Yes	No
	Yes	46	1
	No	50	0

For sexual abuse, the distribution of values found across case file report and child self-report did not significantly differ (χ² (1) = 1.76, p = .186), although interesting patterns of mismatch emerged. Case file records indicated that 59 youth in the sample had no history of sexual abuse; however, 31 (53%) of these 59 children self-reported a history of sexual abuse. Alternatively, of the 38 children who had sexual abuse reported in their file, only 18 (47%) also self-report sexual abuse. Although approximately half of the reports across child self-report and case file record did not match, meaning that some children who self-reported a history of sexual abuse did not have a history of sexual abuse documented in their case file, and vice-versa, overall disagreement was not statistically significant.

For neglect, the assumptions for the McNemar test were not met due to very few cases in which self-report indicated neglect but the case file did not (>5). Thus, a nonparametric binomial test was conducted. Discrepancies between case file and self-report of abuse were also not significant (χ² (1) = 1.57, p = .210). Case file records indicated that 17 youth in the sample had no history of neglect; however, 15 (88%) of these 17 children self-reported neglect. Yet, of the 80 children who had neglect reported in their file, almost all (78, 98%) also self-reported neglect.

Finally, for psychological abuse, results of the McNemar test for psychological abuse indicated that the distribution of values found in youth self-report and case file report significantly differed (χ² (1) = 45.18, p < .001). Case file records indicated that 47 youth in the sample had a history of psychological abuse; however, 96 children, or 98% of the entire sample, self-reported psychological abuse. Almost all of the 47 of children who had psychological abuse reported in their file (46, or 99%) also self-reported psychological abuse. Patterns in the discrepancies between case file and self-report suggested that psychological abuse experiences were identified much more frequently through self-report than case file report.

Next, analyses were conducted to see if age was related to concordance between case file and self-report for all four types of abuse. The sample was split into two groups, one group ages 8 to 12 (n = 38) and the other age 13 to 22 (n = 62). Then, the same series of McNemar tests were conducted except this time per age group. Results for physical abuse (χ² (1) = 12.19, p < .001 for 13 and older and χ² (1) = 4.083, p < .05 for 12 and younger), neglect (χ² (1) = 1.786, p = .18 for the 13 and older group and χ² (1) = 0.00, p = 1.01 for the 12 and younger group), and psychological abuse (χ² (1) = 29.257, p < .001 for the 13 and older group and χ² (1) = 14.062, p < .001 for the 12 and younger group) were the same as the results obtained in the overall analyses. However, results for sexual abuse differed (Table 2). The distribution of values found in youth self-report and case file report significantly differed for the 12 and younger group (χ² (1) = 6.667, p < .01); however, consistent with results of the overall analyses, the distribution of values found across case file report and child self-report did not significantly differ for the 13 and older group (χ² (1) = .000, p = 1.00). Specifically, case file records indicated that 25 youth in the younger group had no history of sexual abuse; however, 10 (40%) of these 25 children self-reported sexual abuse. Then, of the 13 younger children whose case file indicated exposure to sexual abuse, only two of them (15%) did not also self-report a history of sexual abuse (see Table 2), again indicating that overall, more children in the younger group reported some kind of sexual abuse experience than was reported in their case file.

Table 2.

Congruence: Case File Report and Child Self-Report per Age Group

Type of Abuse	Case file report	Self-report
Sexual abuse (13 and older)		Yes	No
	Yes	10	16
	No	15	19
Sexual abuse (12 and younger)		Yes	No
	Yes	13	2
	No	10	12

Finally, a few additional analyses were conducted to see if, across abuse types, younger children more frequently self-reported abuse than older children. Frequencies were obtained to determine how often case files compared to self-reports of abuse indicated a history of all four types of abuse per age group. Results in Table 3 indicate that in general, younger children more frequently self-reported abuse that was not found in their case file than older children.

Table 3.

Congruence Between Case File Report and Child Self-Report

Age	Case file contains all four types of abuse	Self-report contains all four types of abuse
13 and Older	Yes: 6	Yes: 24
	No: 56	No: 36
12 and Younger	Yes: 5	Yes: 20
	No: 33	No: 17

Discussion

Understanding the problem of child maltreatment requires that clear methods be used to document the prevalence of the problem. Without knowing the extent or nature of exposure to child maltreatment, it is fairly impossible for service providers to deliver needed treatment for youth and families. It is important, therefore, for the field and for service professionals to be clear about the source of information on child maltreatment. The present study addressed this need by providing an examination of how reports of abuse differ between the two most common sources of determining maltreatment history, case file and self-report, of four broad categories of abuse: physical, sexual, psychological, and neglect.

Current results corroborate previous findings and suggest that in general, children self-report a history of each type of abuse more frequently than is identified by case file report (Everson et al., 2008; Shaffer et al., 2008), even when substantiated and unsubstantiated reports in the file are included. Also like past research, results indicated that rates of psychological abuse have the least amount of agreement, followed by physical abuse and then sexual abuse, for which there was not a significant overall discrepancy between reports, although the younger children in the study reported significantly more sexual abuse than indicated in their case files (Everson et al., 2008). Results also demonstrated that rates of neglect were similar between case file and youth self-report, a previously understudied area and a new finding for the field.

Because age of the reporter is an important consideration for research, based on concern over a child’s ability to recall events, age was evaluated as a factor related to concordance. Results suggest that the age of the child at the time of the study was a statistically significant factor for the concordance for sexual abuse reports between case file and self-report. Whether statistically significant or not, it also is important to note that younger children (8 to 12) reported more of all four types of abuse than documented in their case file than older children (13 to 22).

Although knowledge of overall and age related differences in rates of each type of abuse found between reports is important, it is just as important to acknowledge that a history of abuse was sometimes found by only one report. It appears that each method may provide a unique contribution in understanding rates of maltreatment. For physical abuse, almost all children whose case file records indicated an allegation of physical abuse also self-reported physical abuse. Perhaps more importantly, for those children whose case file records indicated no allegation of physical abuse, almost all of these children self-reported an incident of physical abuse. This discrepancy could be for several reasons. Previous research suggests that there are variations regarding the training of professionals in recognizing and reporting child physical abuse as well as individual differences regarding the level or severity of physical abuse that warrants a report to CPS (Flaherty et al., 2008; Kellog, 2007). It may be the case that for those children who self-reported physical abuse yet had no documented allegation of physical abuse the abuse the child reported was not shared with CPS or that it was shared but viewed by CPS as an event that did not warrant documentation. As a result, some incidents of physical abuse may go undocumented in case files. Researchers who rely solely on case file report to determine if their sample was physically abused may unknowingly exclude false negatives.

Interesting discrepancies were also found regarding the direction of mismatch for sexual abuse. Although in general there was as much agreement as disagreement between both sources regarding whether a child had a history of sexual abuse, it is particularly interesting that sexual abuse was present in the case file records of many youth who did not self-report being sexually abused – particularly in youth ages 13 and older. Although almost half of the sample did report some kind of sexual abuse event, it is possible that older youth were reluctant to disclose sexual abuse, even over a private computer (Paine & Hansen, 2002), making it understandable that case files might contain reports of abuse that are not identified by self-report. Studies that identify contexts in which children are most likely to report abuse (e.g., with trusted adults, if the abuse becomes increasingly severe) would help elucidate why many children do not self-report sex abuse that is documented in their file even in contexts in which privacy is ensured.

It is also important to consider that a child’s history was coded as including abuse if the child endorsed any instance of sexual abuse. To be included in the yes group for sexual abuse, the child could have endorsed one or all 12 items in the sexual abuse survey. To be coded as a no, the child had to say no to all 12 items. So although the results reported are indicators of overall match or mismatch, they should not be interpreted to mean that the child and the case file, when in agreement, were reporting the same kinds or specific events. Thus, discrepancies between the case file and the child-report are perhaps even more striking and suggest that when research relies solely on the grouping of youth as sexually abused, based on the information in their case file, it is possible that at least half of that sample would not agree, especially perhaps for youth ages 13 and older, or acknowledge that they had ever experienced sexual abuse.

Regarding neglect, case file and child self-report were concordant, and in some ways this is surprising. Although it may be fairly easy for adults outside the family to identify neglect of a child, children may not always be clear about whether or not they experienced neglect, especially if the care is consistent with what they have always experienced. Moreover, although the researchers did not use the technical term “neglect” when assessing for the occurrence of neglectful behaviors as self-reported by children, it is believed that recognizing neglect can equate to knowing that one was not getting their needs met in some way. An example of how these questions were presented to the clients is as follows: “In your lifetime, how often did your parent(s) make sure you bathed regularly?” What bathing regularly means to a child may not be consistent with adult standards. If one’s friends and family do not “bathe regularly,” it may not necessarily be the case that a child would know that their bathing habits are problematic. Because there is little past research examining neglect reporting, the new finding for neglect suggests that youth may be just as aware as adults when their care is sub-par. This concordant (as opposed to the expected discrepant) finding could be due to the ease of observation of situations considered neglectful as opposed to the difficulty identifying more covert types of abuse.

Concordance between case file and self-report of psychological abuse was the most discrepant of the four types examined, with children self-reporting almost twice as many instances of psychological abuse as indicated in the child’s case file. The fact that psychological abuse often coincides with other types of abuse may partially account for the discrepancy. Specifically, reporting the emotional turmoil that may co-occur with sexual abuse, physical abuse, and neglect may not be common, and when psychological abuse co-occurs with other types of abuse, it is likely that the more observable form of abuse (i.e., physical) will be the one to be reported. It also might be unlikely for a case worker or other reporter to witness a child being yelled at, for example, because psychological abuse occurs in the context of interpersonal relationships (O’Hagan, 1995) and does not leave outward “markings,” making it potentially harder to document by outsiders. Further, when psychological abuse is observed, it may still be underreported when the criteria for reporting requires reporters to document harm associated with abuse – which may not be clear when the abuse is primarily psychological (Glaser, 2002).

One of the contributions of the present study was the examination of the possible relation between age and concordance rates between child and case file report. The results suggest that the agreement (or disagreement) between sources of reports of abuse was not dependent on age (except for sexual abuse and younger youth). It is important to note that the present study did not account for when the abuse occurred or how old the child was at the time of the abuse, both factors that could have an impact on what a child recalls. Although current tools for assessing abuse do not allow for asking youth younger than age 8 (the youngest child in this sample), it is possible that memory for events could have degraded over time and that the youth in the present study were no longer able to remember some of the specific events in their history – which may have especially been true for sexual abuse victims, who are likely removed from their homes at a faster rate and following fewer hotline calls (e.g., Bhatti-Sinclair & Sutcliffe, 2012).

Previous research has hypothesized that age may be related to concordance between reports of abuse because children are sometimes less reliable reporters than adults (Amaya-Jackson et al., 2000) and because time since the abuse event could decrease the reliability in self-reports (Greenhoot, 2011). Younger children did more frequently self-report abuse that was not found in their case file than older children, which could be due to a general acquiescent or unreliable reporting style or to better memory for more recent events. However, for three types of maltreatment, age did not impact the discrepancies found between reports. One reason age of the child may have been less of a factor is that this study did not evaluate agreement between specific abuse events but rather agreement regarding whether there was any indication of a history of four different types of abuse. Although the present study did not examine concordance for specific events (i.e., being hit vs. being burned), results suggest that when abuse is considered broadly, age may not be a confounding factor. Another reason why age may not have mattered is because time since the event was not controlled (children did not report on time since the event). It is possible that the sample could have contained a combination of younger children being poorer reporters of abuse (i.e., inconsistent) as well as older children potentially reporting on abuse that occurred several years ago, making age effects difficult to detect.

Age differences aside, it is important for the field to consider the implications of the present results regarding maltreatment reporting discrepancies for youth well-being. It is likely meaningful for service providers to understand that children may know about more maltreatment events than the state authorities or other adults involved in their care. It should be noted that in this study, self-report of abuse and/or neglect that was not found in the case file was not subsequently reported. Although this decision appears contrary to the study’s purpose, it was made due to the impact reporting past abuse might have on the accuracy of child self-report of abuse. Study developers were concerned that some children may not report previous instances of abuse if there were a risk of further reports on their original caregiver. Although the authors ultimately believe that it is important to obtain both self- and case-file reports of abuse when making decisions about treatment – including need for out-of-home placement – past reports of abuse were not reported due to a desire to fully understand the discrepancies between case file and child report. Future researchers should consider reporting past abuse events to CPS and mentioning this reporting as a limit to confidentiality. It would be important to see if such a caveat makes a difference in discrepancies found, or leads to improvements in the child’s care.

Although it may not be germane to treatment to know every instance when the child was mistreated, it is sobering to consider that children and the official record disagree as much as the present results suggest just in regard to type of abuse. It is possible that children are poor reporters and that the disagreement in reports is really the product of the child mis-remembering what they have experienced. It is not clear, however, that the accuracy of reports matters as much as the child’s perception that these events occurred. Moreover, child reports should likely be considered as part of the data gathering process when a child enters into care.

The present results suggest two considerations: (a) professionals should use caution when considering the accuracy of the report from CPS records because for some children, either in reality or just in their minds, other kinds of maltreatment may have occurred, and (b) erring on the side of caution, professionals should be prepared to provide more support than is perhaps warranted by the CPS report of maltreatment. It may be the case that what the state knows is limited or what the child perceives is exaggerated, but in either case, in the name of ensuring the well-being of the child, professionals are encouraged to seek as many avenues for support or intervention as possible so that unknown or unreported experiences might also be remediated and resolved.

Limitations of the Present Study and Directions for Future Research

The findings presented here are important to the field’s understanding of the relation between youth or first-party reporting and case file reports of abuse. Knowing if and what type of abuse has occurred in the life of a child is important for child welfare and child mental health providers in their efforts to determine the best next steps for a child’s well-being. Knowing who to ask is not a simple task, and the results suggest some patterns that could be meaningful for researchers examining rates of abuse and for child welfare providers determining need for services. Although the present findings are important, they are not without limitations.

One, the data in the case files were compiled by several different case workers, some of which likely were responsible for documenting abuse for more than one child in the study. Therefore, there is probably unaccounted for dependence in the data that could have influenced the number and kind of reports available in the child’s case file. Although it was not possible to randomly assign case workers to case files, it is important to consider that the experience level of the case worker could have influenced whether abuse was or was not indicated in the child’s case file. A second limitation is the fact that severity of abuse was not considered. It could have been more likely for the case file and child report to agree for when the abuse was severe (i.e., great physical harm) in nature given the centrality that highly traumatic events sometimes maintain in one’s memory (McNally, 2005). Third, the data collected for the present study were inclusive over the child’s lifetime. It is possible that youth whose abuse events were more recent were better able to recall the events than youth whose abuse experiences were older. A final limitation is that self-report items contained specific examples of types of abuse instead of asking the child broadly whether or not they were maltreated. One implication is that children may have reported the occurrence of life events that they did not consider to be abusive necessarily but still fit the legal definition of abuse. Overall, it appears that when considering the agreement among case file and youth report of abuse for a history of broad categories of abuse, the two reports may indicate similar rates of neglect and sexual abuse (except younger children, who self-report more abuse than what is in their case file) and discrepant reports of physical and psychological abuse.

Future research should focus on the reasons for the discrepancies found between reports. Research could identify not only reasons why particular types of abuse are more likely to be reported by one type of report than another, but also characteristics associated with the reporters that affect reporter reliability. Variables to consider could be child age at the time of the report, time since the abuse event reported, and factors related to the abuse (severity and frequency). Investigation of the relationship between these factors and psychosocial outcomes would also help. It might be important that a child remembers the specifics of their neglectful experiences, such as severity and frequency, and that this ability to specifically recall details surrounding one’s abuse is related to psychosocial outcome. Yet, it might also be that a child simply being able to indicate that they were neglected, no matter the frequently or severely, is most important when determining outcomes. Although understanding agreement between reporters regarding severity and frequency could be important, it may not provide additional predictive value when determining abuse-related outcomes. Future research could create guidance for researchers and clinicians when selecting methods to determine abuse outcomes and service provision.