Recent Trends in Local Child Protective Services Practices

Contents

Executive Summary

Chapters

1. Introduction
2. Changes in CPS Structure and practice
   1. Agency Administration and Staffing
   2. CPS Practices in Screening/Intake, Investigation, and Alternative Response
   3. Collaboration with Other Agencies
3. Relationships between Local CPS Agency Characteristics and Child Maltreatment in the Agency Jurisdiction
   1. Rate of Alleged Maltreatment
   2. Percentage of Children Assigned a Substantiated or Indicated Disposition
   3. Rate of Physical Abuse Alone
   4. Rate of Neglect Alone
   5. Rate of Sexual Abuse Alone
   6. Rate of Psychological Maltreatment Alone
   7. Rate of Medical Neglect Alone
   8. Rate of Multiple Maltreatment
4. Conclusion

References

List of Tables

Tables in Chapter 2:

1. Functions Provided by Local CPS Agencies
2. Assignment across Screening/Intake and Alternative Response
3. Excessive Workload Concerns
4. The Most Common Referral Sources
5. Acceptance of Referrals During Non-business Hours
6. Acceptance of Referrals from Non-English Speakers
7. Required Standard Practices  Investigation
8. Required Standard Practices  Alternative Response
9. Activities Always Conducted During an Investigation
10. Activities Always Conducted During Alternative Response
11. Instruments and Tools  Alternative Response
12. Always Present an Obstacle to Timely Completion of Investigations
13. Always Present an Obstacle to Timely Completion of Alternative Response
14. Services Provided by CPS Agencies after Completion of Investigation
15. Sole Responsibility for Investigating Specific Maltreatment Types
16. Sole Responsibility for Alternative Response by Maltreatment Type

Tables in Chapter 3:

1. Agency Characteristics Examined in Relation to Maltreatment Measures
2. Significant Relationships Between Agency Characteristics and the Alleged Maltreatment Rate
3. Significant Relationships Between Agency Characteristics and the Percentage of Children in Investigated Reports Who Were Substantiated or Indicated
4. Significant Relationships Between Agency Characteristics and the Rate of Physical Abuse Alone
5. Significant Relationships Between Agency Characteristics and the Rate of Neglect Alone
6. Significant Relationships Between Agency Characteristics and the Rate of Sexual Abuse Alone
7. Significant Relationships Between Agency Characteristics and the Rate of Psychological Maltreatment Alone
8. Significant Relationships Between Agency Characteristics and the Rate of Medical Neglect Alone
9. Significant Relationships Between Agency Characteristics and the Rate of Multiple Maltreatment

List of Figures

Figures in Chapter 3:

1. Relationship Between the Percentage of Agencies that Reviewed Prior CPS Records for All Cases and the Rate of Alleged Maltreatment
2. Relationship Between the Percentage of Agencies With Sole Responsibility for Investigating Fatalities and the Rate of Alleged Maltreatment
3. Relationship Between the Percentage of Agencies That Had Sole Responsibility for Investigating Severe Physical Abuse and the Percentage of Children in Investigated Reports Who Where Assigned A Substantiated or Indicated Disposition
4. Relationship Between the Percentage of Agencies That Used a State Hotline to Handle Calls on Weekday Evenings and the Percentage of Children in Investigated Reports Who Were Assigned a Substantiated or Indicated Disposition
5. Relationship Between the Percentage of Agencies That Used a State Hotline to Handle Calls on Weekends and the Rate of Physical Abuse Alone
6. Relationship Between the Percentage of Agencies That Provided Homemaker/Chore Services and the Rate of Physical Abuse Alone
7. Relationship Between the Percentage of Agencies That Provided Alternative Response and the Rate of Neglect Alone
8. Relationship Between the Percentage of Agencies Where Having Insufficient Time to Make a Good Determination Was Always an Obstacle to Timely and Accurate Investigations and the Rate of Neglect Alone
9. Relationship Between the Percentage of Agencies Where Investigating Workers Always Interviewed Professionals Known to the Family and the Rate of Neglect Alone
10. Relationship Between the Percentage of Agencies Using a State Hotline to Take Referrals on Weekends and the Rate of Sexual Abuse Alone
11. Relationship Between the Percentage of Agencies That Provided Homemaker/Chore Services After Investigation and the Rate of Sexual Abuse Alone
12. Relationship Between the Percentage of Agencies Reporting That Insufficient Time with the Family Was Always an Obstacle to Completing Investigations and the Rate of Psychological Maltreatment Alone
13. Relationship Between the Percentage of Agencies Reporting an Excessive Investigation Workload and the Rate of Psychological Maltreatment Alone
14. Relationship Between the Percentage of Agencies Where Investigators Discussed All Cases With Other CPS Workers and the Rate of Medical Neglect Alone
15. Relationship Between the Percentage of Agencies That Provided an Alternative Response and the Rate of Multiple Maltreatment
16. Relationship Between the Percentage of Agencies Where Investigating Workers Reviewed Prior CPS Records for All Cases and the Rate of Multiple Maltreatment

Appendices

1. Questionnaire Used in the Local Agency Survey (LAS) in 2002 (in PDF format, 87 pages)
2. Questionnaire Used in the CPS Structure and Practices Mail Survey (SPM) in 2005-2006 (in PDF format, 28 pages)
3. Weighting and Analysis (in PDF format, 4 pages)
4. Comparisons of LAS and SPM Estimates (in PDF format, 10 pages)

Executive Summary

This study examines changes in policies and procedures in local child protective services (CPS) agencies between 2002 and 2005-2006.  The baseline information comes from the 2002 National Study of Child Protective Services Systems and Reform Efforts (U.S. Department of Health and Human Services, 2003).  The 2005-2006 data come from the CPS Structure and Practices Mail Survey (Li, Shusterman & Sedlak, 2009), which is a sub-study of the Fourth National Incidence Study on Child Abuse and Neglect (NIS-4).  Both studies surveyed administrators of nationally representative samples of CPS agencies and used nearly identical survey instruments to collect information about local agencies investigative practices. This analysis is intended to identify noteworthy changes or development in recent years in each of four topical areas covered by the surveys: administration and staffing; screening and intake procedures through which child maltreatment reports are accepted; investigation of child maltreatment allegations; and alternative CPS response.  Alternative responses were defined as any approach the agency has for responding to maltreatment allegations that is not intended to result in a determination about whether maltreatment occurred.  The study also sought to determine whether changes in practice or policy were associated with changes in maltreatment reporting or victimization.

Recent Trends. Despite a short time interval between the two surveys (three to four years), the analysis identified a number of changes in agency operations.  In 2005-2006 as compared with 2002:

• Fewer agencies reported conducting alternative CPS responses (39% in 2005-2006 versus 69% in 2002).
• Fewer agencies expressed concern over excessive investigation workloads (47% versus 64%).
• More agencies reported schools and law enforcement as their most common referral source 43% and 23% respectively in 2005-2006, versus 29% and 7% in 2002).
• More agencies relied on state hotlines to screen CPS referrals during non-business hours.
• Agencies reported an increased number of activities that are required in every investigation.
• Agencies more often reported a range of obstacles to the timely completion of investigations.
• Agencies more frequently offered parenting classes and substance abuse treatment services (96% and 92% respectively in 2005-2006 versus 88% and 85% in 2002).
• Agencies less frequently offered all other types of services including various types of counseling services, homemaker services, and employment services, among others.  Substantial declines were reported in the availability of many services.
• Agencies were less likely to bear sole responsibility for investigating severe maltreatment, including severe physical abuse (7% in 2005-2006 versus 23% in 2002), severe neglect (34% versus 51%), and child fatalities (5% versus 12%).

Associations with maltreatment measures. Through a series of regression analyses, this study examined whether reported changes in agency practices were associated with changes in rates of investigation, rates at which reports were substantiated/indicated, and child maltreatment rates in the local agencies jurisdictions. Relationships were examined between each of fifty agency characteristics described in the survey and eight maltreatment measures: the percentage of investigated children whose maltreatment was substantiated or indicated, the rate of maltreatment reports and investigations in the jurisdiction, as well as the rate of substantiated reports with physical abuse alone, neglect alone, medical neglect alone, sexual abuse alone, psychological abuse alone, and multiple maltreatment types.  Fully one-half of the agency characteristics were not related to any maltreatment measure, and most of the remaining characteristics were related to a single maltreatment measure.  However, four agency characteristics were more widely associated with the maltreatment measures examined.  These were:  (1) whether the agency provided an alternative response option; (2) whether it handled referrals through a state hotline in the evenings; (3) whether it relied on a state hotline on weekends; and (4) whether the agencys investigative workers always reviewed prior child protective services reports when conducting an investigation.  The following relationships were found between these characteristics and maltreatment measures:

• Alternative Response.  Agencies that provided an alternative response had significantly lower rates of neglect, medical neglect, and cases with multiple forms of maltreatment in their jurisdictions. Alternative responses are typically non-investigative assessments of family strengths and needs that allow CPS agencies and their community partners to intervene with families in more supportive ways. Alternative responses are generally provided in cases of lower risk and may also be referred to as  differential responses, multi-track responses, and dual-track responses.
• Hotline for Coverage during Non-Business Hours.  Agencies that used a state hotline to handle calls during non-business hours substantiated or indicated a significantly lower percentage of children in their investigative reports and had significantly lower rates of confirmed physical abuse or sexual abuse in their jurisdictions. However, these agencies also had significantly higher rates of confirmed cases with multiple forms of maltreatment.
• Review of CPS History.  Agencies that always reviewed prior child protective services records during investigations had higher rates of maltreatment in their jurisdictions on a number of measures including the overall rate of alleged maltreatment reports to CPS and rates of sexual abuse, neglect, medical neglect, and cases with multiple forms of maltreatment.

Discussion. Fewer agencies reported offering an alternative, non-investigative response to child maltreatment in 2005-2006 than did just a few years earlier.  Yet those offering an alternative response reported more standardization in their practices, with more activities conducted all or most of the time. This may reflect a better understanding of what alternative response programs are and more accurate reporting of whether agencies practices fit this model. Findings also reflect more standardization of investigative practices, with proportionally more agencies in the later period indicating that their workers always conducted specific activities during an investigation. It also appears that the screening/intake function became more centralized, with fewer local CPS agencies conducting their own screening/intake and more relying on state hotlines, especially during non-business hours. These trends are consistent with state program improvement plans (PIPs) developed in response to weaknesses identified in the federal child welfare monitoring process called the Child and Family Services Reviews (or CFSRs).  State PIPs often include safety strategies focused on improved practice models and increased consistency in investigative practices. With this increased standardization and additional required elements of each investigation, more agencies in the later time period reported facing obstacles to the timely completion of investigations. The increased awareness of barriers to timely investigations may reflect the inclusion of timeliness in CFSR measures of safety outcomes.  Finally, it is encouraging that, while still substantial, the percentage reporting excessive investigative workloads decreased between 2002 and 2005-2006.

Limitations. The findings regarding recent trends in CPS practices are solely descriptive.  However, they may suggest how, if a given relationship and other influencing factors remain stable, substantial and significant changes in the prevalence of certain agency characteristics or practices may relate to changes in maltreatment measures. The findings on the relationship between agency characteristics and maltreatment measures are qualified by several limitations.  First, the level of maltreatment a CPS agency encounters and substantiates or indicates is undoubtedly the result of many factors. These analyses focus on a single factor at a time and do not attempt to control for other influences. Second, the relationship computed from the pooled database may not be as strong in either study individually as it appears in these analyses that combine data, and may or may not be stable in future years. Third, the relationships documented show correlation only and do not indicate causation. For instance, whether an agency has a characteristic or employs a certain practice may affect the number of cases it is able to address; the number of cases an agency encounters may lead it to adopt a certain practice or structure; or both the agencys characteristic and the maltreatment rates observed may derive from a third underlying factor.

Conclusions. This study examined how the delivery of child protective services in the U.S. has changed in recent years. Studies like this one that document changing policies and practices over time provide an important context for understanding performance.  Findings show changes to a number of child protective services practices, resulting in more standardization. At the same time, the availability of services for families has generally declined, though two services, parenting training and substance abuse treatment, became somewhat more available through CPS agencies. Most practice changes appeared unrelated to child maltreatment outcomes over this time period, although alternative response practices were associated with lower maltreatment rates and both reliance on state hotlines for off-hours screening/intake and child abuse history checks were associated with higher maltreatment rates. The decline in service provision is of concern, although in the absence of a strong evidence base regarding service effectiveness, it is hard to judge the impact of service delivery changes. Perhaps the recent emphasis on evidence-based practices in the child welfare field will provide future researchers with clearer links between practices and outcomes for children.

[ Go to Contents ]

Introduction

This study uses two extant surveys to examine changes in policies and procedures in local child protective services (CPS) between 2002 and 2006. The baseline information comes from the Local Agency Survey (LAS) in the 2002 National Study of Child Protective Services Systems and Reform Efforts (U.S. Department of Health and Human Services, 2003). The LAS was administered to a national sample of 375 CPS agencies. Eighty percent of the agencies responded to the survey. The LAS focused on the structure, core functions, and standard practices of the local CPS agencies. It represented a pioneer effort in using data collected from a nationally representative sample to understand the operations of CPS agencies in different domains.

The 2006 data come from the CPS Structure and Practices Mail Survey, the SPM (Li, Shusterman & Sedlak, 2009), which is a sub-study of the Fourth National Incidence Study of Child Abuse and Neglect (NIS-4). The SPM questionnaire was nearly identical to that used in the LAS. The survey was administered to all NIS-4 agencies, a national sample of 126 agencies, collecting information about how the agencies operated at the time of the NIS-4 reference period (either fall 2005 or spring 2006). The response rate for the NIS-4 SPM survey was 98%.

The survey instruments used a modular format, which allowed different sections to be completed by different personnel, depending on who could provide the best information about the module topic. These modules focused on four areas of CPS operation: administrative structure, screening/intake, investigation, and alternative CPS response.  There were slight differences in the instruments used in the two surveys.[1] Appendix A provides the LAS instrument; Appendix B provides the SPM instrument. In each survey, the Administration Module collected information on the basic organization of the agency; the Screening/Intake Module gathered data on the agencys screening practices; the Investigation Module asked about the agencys procedures for conducting investigations of alleged maltreatment; and the Alternative CPS Response Module enabled the agency to describe any alternative approaches it had for responding to allegations of maltreatment other than traditional investigation.

The data records from each survey are weighted, which allows each sample to provide estimates of percentages of local CPS agencies that followed a particular practice during the survey timeframe. The methodologies for the earlier LAS and for the more recent SPM are documented in the respective reports for those surveys (U.S. Department of Health and Human Services, 2003; Li, Shusterman & Sedlak, 2009).

The current analyses aimed to identify noteworthy changes or developments from the LAS to the SPM in CPS operations in and across each of the four topical areas. The analyses examined how local agency practices changed and further explored whether any of the observed changes were related to child maltreatment reports to CPS and/or the incidence of victimization according to CPS records. Specifically, it addressed the following four research questions:

• Have there been noteworthy changes or developments in local CPS practices in the years since the 2002 National Study of Child Protective Services Systems and Reform Efforts?
• What, if any, practice trends can be seen in the areas of agency administration and staffing, screening and intake procedures, investigation and alternative response, and collaborative efforts with other agencies?
• How much change do agencies report overall?
• Are observed structure and practice changes associated with differences in either the incidence of maltreatment reports or the rate of substantiated/indicated victimization?

To identify significant changes from the LAS to the SPM, analysts assessed the statistical significance of differences between the two surveys estimates of the percentages of agencies with particular organizational and operational characteristics in each of the four areas of CPS structure and practices. The following sections present findings in the areas of CPS structure and operation where the LAS and the SPM differed significantly.

In addition, agency characteristics that showed significant changes between 2002 and 2005-2006 were examined in relation to the NCANDS data on the rates of child maltreatment seen by CPS agencies during the two time periods.  Chapter 3 reports findings on the significant relationships between agency characteristics and rates of child maltreatment.

Appendix C describes the special procedures needed to weight the SPM data, apply NCANDS data to compute the maltreatment rates for the LAS and SPM local agencies, and pool the two studies data for the Chapter 3 analyses.  Appendix D provides the results of all the statistical comparisons reported in Chapter 2, which describes differences in the percentages of agencies with specific characteristics in the two studies.

It should be noted that many estimates for the LAS given in Chapter 2 here are close (but not identical) to the estimates provided in the original LAS report. Most discrepancies were due to different methods of handling missing data. In the LAS, cases with missing data were included in calculating percentages, and the LAS tables reported the percentage of agencies with missing data. In this report, those cases were always excluded from the analysis. Footnotes in Chapter 2 explicitly discuss any discrepancies between the original LAS report and the LAS results given here that do not derive from differences in handling missing data.

Changes in CPS Structure and practice

Agency Administration and Staffing

Table 2-1 shows percentage of agencies that provided local screening and intake, investigation, and alternative CPS response.

Eight-five percent of agencies in the LAS and 74% of agencies in the SPM conducted screening/intake at the local level. This difference is statistically marginal, meaning that the estimated change closely approaches statistical significance (i.e. p.10), but it does not meet the traditional statistical standard for concluding that the difference is not due to chance factors (i.e., p.05).

All local CPS agencies in both surveys indicated that they conduct investigations on referred cases; the prevalence of this CPS function did not change between 2002 and 2006.

Sixty-nine percent of agencies in the LAS provided alternative response, compared to only 39% of agencies providing alternative response in the SPM. This difference is statistically significant.

A smaller percentage of agencies assigned entirely different (i.e., specialized) workers to screening/intake and alternative response in 2005-2006 at the time of the SPM than in 2002 at the time of the LAS (Table 2-2). Compared to agencies practices in 2002, agencies in 2005-2006 were proportionally more likely to have workers who routinely conduct both screening/intake and alternative response (71% vs. 42%).

Proportionally, fewer agencies in the SPM said they experienced excessive workload in investigation or in alternative response than did agencies in the LAS (Table 2-3). Nearly two-thirds (64%) of agencies said they experienced excessive investigation workload in the LAS, compared to less than one-half (47%) of agencies in the SPM, a statistically significant decrease. Among agencies that conducted alternative response, the percentage that claimed they experienced excessive workload in this function decreased between the LAS and the SPM (42% vs. 23%). This decrease is statistically marginal. There was no change in the percentage of agencies that experienced excessive workload in conducting screening/intake.

CPS Practices in Screening/Intake, Investigation and Alternative Response

Both the LAS and the SPM asked agencies to rank the referral sources from which they received reports of child maltreatment.  Table 2-4 shows the percentages of agencies that ranked school, law enforcement, individuals (general public), or hospital/health professionals as their most common referral source. Referral sources that did not appear in both survey questionnaires are omitted from the table.

Twenty-three percent of agencies in the SPM ranked law enforcement as their most common referral source, whereas only 7% of agencies in the LAS did so. The increase of 16% from the LAS to the SPM is statistically significant. The percentage of agencies that ranked schools as the most common referral source increased by 14% from the LAS to the SPM (29% vs. 43%). This difference is also statistically significant. In contrast, the percentage of agencies that ranked individuals as the most common referral source decreased from 33% to 23% during the same period, a statistically marginal decrease. There was no statistically reliable change in the percentage of agencies that ranked hospitals/health professionals as their most common referral source.

Some significant differences emerged in the ways agencies handled referrals during non-business hours (Table 2-5). During weekday evenings, 18% of agencies at the time of the LAS (2002) but 41% of agencies at the time of the SPM (2005-2006) relied on state hotlines to handle referrals, a statistically significant increase. In addition, fewer agencies at the time of the SPM routed referrals to another agency (1% vs. 5%) or handled them using another method not specified in the questionnaire (6% vs. 19%). All of these differences are statistically significant. Agencies in the SPM were also proportionally less likely to assign referrals received during weekday evenings to on-call staff (41% vs. 54%), a statistically marginal difference.

Similar patterns emerged in how agencies handled referrals on weekends. A significantly higher percentage of agencies at the time of the SPM (2005-2006) relied on state hotlines to handle the referrals (41% vs. 17%). In contrast, the agencies in the SPM were less likely to route these referrals to another agency (1% vs. 5%) or to handle them using another method (6% vs. 20%). All of these differences are statistically significant. Another difference between the two time periods emerged as statistically marginal: in 2005-2006 agencies were more likely than in 2002 to handle the referrals using their own intake unit (7% vs. 2%).

Table 2-6 shows how CPS agencies handled referrals from non-English speakers. Whereas 13% of agencies in the 2002 LAS used non-English speakers on staff to handle these calls, only 3% of agencies in the 2005-2006 SPM used this method. This decrease is statistically significant. The LAS and SPM also differed in the percentage of agencies that employed multiple methods to handle referrals from non-English speakers. Only 17% of agencies employed multiple methods for these referrals in 2002 at the time of the LAS, compared to 38% of agencies in 2005-2006, at the time of the SPM, a statistically significant increase.

More agencies indicated that specific standard practices were required when conducting investigation in 2005-2006 at the time of the SPM than in 2002 at the time of the LAS (Table 2-7). Compared to agencies practices in 2002, agencies in 2005-2006 were proportionally more likely to make a determination of whether one or more children were at risk of maltreatment (99% vs. 88%), make a determination on all children in the family as to whether they had been maltreated (98% vs. 81%), make a recommendation for court intervention if needed (92% vs. 83%), refer the family for further services if needed (79% vs. 68%), or provide short-term services if needed (76% vs. 60%). All of these differences are statistically significant. Further, agencies in the SPM were more likely than those in the LAS to make an assessment of service needs of the child (92% vs. 86%). This difference is statistically marginal.  There were no statistically reliable differences in the percentages of agencies that would make an assessment of the familys immediate service needs, remove the child if immediate safety is an issue, or make a determination as to whether the referred child had been maltreated.

The analyses also revealed differences in the percentages of agencies that required specific standard practices when conducting alternative response (Table 2-8). In 2005-2006, at the time of the SPM, the large majority of agencies that conducted an alternative response (84%) said that they required as standard practice that the worker assess the underlying causes of the maltreatment incident, compared to just a slight majority (51%) of agencies in 2002 at the time of the LAS. This difference is statistically significant. Agencies in the SPM were also proportionally more likely than those in the LAS to require that the caseworker make an assessment of the service needs of the entire family (86% vs. 69%), make an assessment of the service needs of the child (86% vs. 70%), and refer the family for further services if needed (71% vs. 51%). These differences are statistically marginal. There were no statistically significant or statistically marginal differences in the percentages of agencies that required the worker to provide short-term services or to make a recommendation for court intervention if needed.

Forty-six percent of agencies in 2005-2006 at the time of the SPM always notified the person who reported the alleged maltreatment when a determination was made that the maltreatment had occurred. This is a statistically significant increase from the prevalence of this practice in 2002 at the time of the LAS, when only 30% of agencies did so. The differences in the percentages of agencies that would always notify the perpetrator or always enter the perpetrator in a central registry at the conclusion of an investigation were not statistically significant.

Several significant changes were evident in the activities that agencies always conducted during investigations (Table 2-9). When conducting a CPS investigation, higher percentages of agencies in 2005-2006 at the time of the SPM always reviewed prior CPS records (97% vs. 90%), interviewed family members other than caregivers (86% vs. 56%), visited the family without an appointment (72% vs. 35%), interviewed professionals known to the family (64% vs. 41%), discussed the case with other CPS workers (60% vs. 24%), conducted criminal background checks on alleged perpetrators (48% vs. 32%), visited the family with an appointment (42% vs. 6%), or conducted a family group conference meeting (33% vs. 6%). The percentage of agencies that always interviewed witnesses was marginally higher during the later SPM period than during the earlier LAS timeframe (77% vs. 67%). In addition, a marginally higher percentage of agencies always interviewed the reporter at the time of the SPM (54% vs. 44%). During the earlier timeframe, however, more agencies always discussed the case with a multidisciplinary team (10% vs. 5%). There were no statistically reliable differences in the percentages of agencies that would always obtain/preserve physical evidence, remove the child harmed or in danger of harm, interview or formally observe the child, or interview caregiver(s).

Table 2-10 shows activities always conducted during the agencys alternative (non-investigative) response. In providing alternative response, significantly more agencies during the later SPM timeframe (2005-2006) always discussed the case with other CPS workers (50% vs. 19%), and marginally higher percentages of agencies always interviewed the caregiver(s) (87% vs. 74%), interviewed or formally observed the child(ren) (85% vs. 71%), or visited the family with an appointment (34% vs. 15%). There were no statistically reliable differences in the percentages of agencies that would conduct a family group conference meeting, visit the family without an appointment, interview the person who reported the alleged maltreatment, conduct criminal background check on the alleged perpetrator, or interview professionals known to the family. The analyses could not assess the reliability of the change in the percentage of agencies that always discussed the alternative response case with a multidisciplinary team, because the percentage of agencies doing so was too small to be estimated on the basis of the SPM sample.

Analyses examined but found no statistically reliable differences in the percentages of agencies that used any of the listed tools and instruments during investigation.  These included: a structured decision-making model, any formal instrument for assessing safety or risk, or any guideline for establishing risk or safety, and any standardized instrument for assessing substance abuse, domestic violence, parenting skills, child development, or family support.

Regarding use of various instruments and tools during alternative response, the analyses revealed that, compared to practices at the time of the 2002 LAS, a significantly lower percentage of agencies during the 2005-2006 SPM used a standardized domestic violence assessment instrument (3% vs. 11%) in their alternative response (Table 2-11). In addition, a marginally lower percentage of agencies used a formal risk assessment instrument when conducting alternative response during the latter time period (17% vs. 31%).  There were no statistically reliable changes in the percentages of agencies that used standardized child development inventory, standardized substance abuse assessment instrument, formal safety assessment instrument, structured decision-making model, or guideline for establishing risk or safety. The analyses could not assess the reliability of the change in the percentage of agencies that used a standardized parenting skills assessment or a standardized family support assessment, because the percentage of agencies doing so was too small to be estimated on the basis of the SPM sample.

The survey questionnaires asked respondents to identify factors that presented problems for their agency in completing an investigation or an alternative response in a timely manner. Table 2-12 shows the percentages of agencies that said the activity in question always posed an obstacle to their timely completion of an investigation.

In 2005-2006, more agencies identified each of the activities listed in the table as an obstacle to their timely completion of investigations. The largest increases from the time of the 2002 LAS occurred in explaining the consequences of actions to parent(s) (52% vs. 3%), handling language barriers (45% vs. 2%), having sufficient time to make good determination (34% vs. 12%), and assessing parenting skills (33% vs. 2%). The differences listed in Table 2-12 are all statistically significant. There was no statistically reliable change in the percentage of agencies that said obtaining necessary expertise from other professionals was always an obstacle to completing an investigation;  there were too few sample cases to assess change in the percentage of agencies that claimed that locating the family was always and obstacle.

Table 2-13 gives the percentages of agencies that said the activity in question always posed an obstacle to their timely completion of alternative response.  Readers should note that because of the low number of SPM sample agencies that conducted an alternative response and the low percentages of LAS agencies that reported these obstacles, nearly all estimates in this table are based on low numbers of sample agencies and so are less reliable.  Similar to the results on obstacles for completing investigation, whenever analyses detected a statistically reliable change since the LAS in 2002, more agencies identified the activity as always posing an obstacle in 2005-2006 at the time of the SPM.  Compared to the LAS, significantly higher percentages of agencies in the SPM said that the following activities always posed obstacles to their timely completion of alternative responses: preparing materials for the case record (58% vs. 9%), handling language barriers (50% vs. 4%), explaining the consequences of actions to parent(s) (40% vs. 7%), and obtaining necessary expertise from other professionals (23% vs. 4%). . In addition, higher percentages of agencies at the time of the SPM said they always experienced obstacles to completing alternative response in a timely manner related to determining what happened to the child (17% vs. 4%), locating the family (18% vs. 1%), and assessing parenting skills (14% vs. 2%). These last three percentage differences are statistically marginal.

The prevalence of post-investigation services changed between the time of the LAS and that of the SPM, but different services changed in different directions (Table 2-14). Compared to the LAS 2002 baseline, agencies in the 2005-2006 SPM were significantly more likely to provide parenting classes (96% vs. 88%) and substance abuse programs (92% vs. 85%). However, fewer agencies provided the other services listed at the time of the SPM.  During the later timeframe, significantly fewer agencies provided marital counseling (51% vs. 71%), family systems therapy (52% vs. 74%), grief counseling (49% vs. 68%), advocacy services (45% vs. 69%), dental exams (43% vs. 58%), homemaker/chore services (39% vs. 63%), or financial planning services (24% vs. 57%). Marginally fewer agencies provided employment services (47% vs. 58%).

Collaboration with Other Agencies

Agencies indicated their roles in conducting investigation and alternative response by indicating whether, for different types of cases, they had sole responsibility, shared responsibility with law enforcement, and/or shared responsibility with other, non-law enforcement agencies. They indicated their roles for cases involving different types of maltreatment and different categories of perpetrators.

The two surveys showed that there were significant changes in the percentages of agencies with sole responsibility for investigating several types of maltreatment (Table 2-15). Agencies in the 2005-2006 SPM were proportionally less likely than those in the 2002 LAS to have sole responsibility for investigating the more serious forms of maltreatment, including severe physical abuse (7% vs. 23%), severe physical neglect (34% vs. 51%), and child fatality (5% vs. 12%). They were more likely than agencies in the LAS to have sole responsibility for investigating truancy (12% vs. 6%). There was no statistical reliable difference in percentages of agencies that had sole responsibility for investigating abandonment. The analysis could not examine agencies responsibility for investigating several forms of child maltreatment (including sexual abuse, other physical abuse, other physical neglect, and emotional abuse and neglect) because the LAS and SPM questions about these maltreatment forms were not the same.

The 2005-2006 SPM showed that fewer agencies held sole responsibility for alternative response across all types of maltreatment listed in Table 2-16. Statistically significant differences between the 2005-2006 SPM and the 2002 LAS occured in the percentages of agencies with sole responsibility for the following maltreatment categories: severe physical abuse (2% vs. 14%), severe physical neglect (12% vs. 25), abandonment (14% vs. 34%), and child fatality (1% vs. 8%). The difference in the percentage of agencies that had sole responsibility in alternative response for truancy was neither statistically significant nor statistically marginal. Once again, the analysis could not examine differences in responsibility for sexual abuse, other physical abuse, other physical neglect, and emotional abuse and neglect, because the two surveys asked different questions about agencies responsibility for these forms of child maltreatment.

Nineteen percent of agencies in 2005-2006 at the time of the SPM, but just 8% of agencies in 2002 at time of the LAS, had sole responsibility for investigating child maltreatment perpetrated by a person who was not a caregiver. The difference is statistically marginal.

In addition, the 2002 LAS and the 2005-2006 SPM differed significantly in the percentage of agencies that shared responsibility with other agencies for investigating child maltreatment perpetrated by foster parent. Sixteen percent of agencies in the LAS, but 30% in the SPM, shared this responsibility with other agencies. No other difference between the 2002 LAS and the 2005-2006 SPM emerged as statistically significant in shared responsibility for investigation related to the type of perpetrator.

[ Go to Contents ]

Relationships between Local CPS Agency Characteristics and Child Maltreatment in the Agency Jurisdiction

A series of regression analyses examined the relationships between differences in CPS structure and practices, as assessed in the LAS and SPM studies, and rates of child maltreatment derived from NCANDS data. These analyses used the local CPS agency as the unit of analysis and focused on those aspects of local CPS structure and practice that evidenced changes between 2002 and 2005-2006, as described in the previous chapter.  The analyses related these features of local CPS structure and practice to rates of investigation, substantiation, and child maltreatment in the local agencies jurisdiction.  For the most part, an agencys jurisdiction was an entire county and most agencies served just a single county.[4]

The measures of rates of investigation, substantiation, and child maltreatment were derived for each local agency in the LAS and SPM samples from 2002 and 2006 case-level maltreatment data provided to the National Child Abuse and Neglect Data System, NCANDS (U.S. Department of Health and Human Services, 2008). Specifically, the analyses examined the following measures for the study year in question:[5]

• Number of children alleged to be maltreated on investigated CPS reports per 1,000 children residing in the agencys jurisdiction
• Percentage of investigated children whose maltreatment the CPS agency substantiated or indicated
• Number of children per 1,000 residing in the agencys jurisdiction that CPS substantiated or indicated with physical abuse alone (i.e., and no other maltreatment)
• Number of children per 1,000 residing in the agencys jurisdiction that CPS substantiated or indicated with neglect alone (i.e., and no other maltreatment)
• Number of children per 1,000 residing in the agencys jurisdiction that CPS substantiated or indicated with medical neglect alone (i.e., and no other maltreatment)
• Number of children per 1,000 residing in the agencys jurisdiction that CPS substantiated or indicated with sexual abuse alone (i.e., and no other maltreatment)
• Number of children per 1,000 residing in the agencys jurisdiction that CPS substantiated or indicated with psychological maltreatment alone (i.e., and no other maltreatment)
• Number of children per 1,000 residing in the agencys jurisdiction that CPS substantiated or indicated with multiple maltreatment types

In preparation for these analyses, the LAS agency records were merged with the 2002 NCANDS maltreatment information on the jurisdictions, and the SPM records were merged with the 2006 NCANDS maltreatment measures. Appendix C provides details about this merge and the adjustments needed to ensure correct variances in the significance tests and to minimize potential bias from the lack of NCANDS data for some jurisdictions.

The analyses examined the relationship between each of the 50 agency characteristics in Table 3-1 and each of the 8 maltreatment measures, described above.  These are all the agency features that demonstrated significant differences between the LAS and the SPM (as reported in Chapter 2), excluding agency differences in conducting alternative response[6] or where data were insufficient to assess the relationship.[7] These were used as independent variables in bivariate linear regression models, examining their relationship to each of the maltreatment measures listed above.  Because of the number of regressions involved, this chapter ignores marginal results and presents only those relationships that meet the traditional standard for statistical significance (i.e., p.05). Each maltreatment measure was significantly related to only a minority of agency characteristics. The tables below list these agency characteristics, their regression coefficients, and the t score for each significant regression model.

One-half of the agency characteristics these analyses examined (25) were not significantly related to any maltreatment measure. Table 3-1 indicates these at the end of their respective lists, flagged with an asterisk. Of the remaining 25 agency characteristics examined here, most related to just one or two maltreatment measures.  However, four agency characteristics had a more pervasive connection to maltreatment indictors, having significant associations with three or more of the maltreatment measures.  These were (1) whether the agency provided an alternative response option, (2) whether it handled referrals through a state hotline in the evenings, (2) whether it relied on a state hotline on weekends, and (3) whether the agencys investigative workers always reviewed prior CPS records.  The sections below provide details on these findings.

The graphs in this chapter depict the relationships between the percentage of agencies with a given structure or practice and the prevalence of child maltreatment or the percentages of children assigned a substantiated or indicated disposition. In each graph, the percentage of agencies with the characteristic in question is on the X-axis, while the maltreatment measure is on the Y-axis.  The slope of the regression line indicates whether the relationship between the agency characteristic and the maltreatment measure is statistically positive or negative. The relationship is statistically positive if the slope rises from left to right, meaning that agencies with the characteristic have higher values on the maltreatment measure in their jurisdictions. It is statistically negative if the slope falls from left to right, meaning that agencies with the characteristic have significantly lower values on the maltreatment measure.

In addition to slope lines, each graph also depicts two points (represented by stars), which reflect the overall values of the agency and maltreatment measures in each study, the 2002 LAS and the 2005-2006 SPM. The value on the X-axis corresponds to the percentage of CPS agencies with the characteristic in question during the study year.  These are the same percentages provided earlier in Chapter 2.  The Y-axis value for each point is the national estimated rate of the child maltreatment measure, obtained for each study year, based on all the local-agency records with available information in the pooled database.[8]

The tables below report 42 significant relationships between local agency characteristics and the maltreatment measures in their jurisdictions.  When considering the findings reported here, readers should bear three cautions in mind:  First, that the level of maltreatment a CPS agency encounters and substantiates or indicates is undoubtedly the result of many factors. These analyses focus on a single factor at a time and do not attempt to control for other influences.  Second, the relationship computed from the pooled database may not be as strong in either component study and may or may not be stable in future years.  Third, although the terminology of linear regression labels one component the independent variable and another variable the dependent variable, the analyses simply document the correlation of these variables.  Correlation does not indicate causation. Whether an agency has a characteristic or employs a certain practice may affect the number of cases it is able to address;  the number of cases an agency encounters may lead it to adopt a certain practice or structure; or both the agencys characteristic and the maltreatment rates observed here may derive from a third underlying factor (e.g., size or population density in the jurisdiction).

For the large majority of the relationships reported in this chapter, the national measures for each study (shown by the starred points in the graphs) align with the slope of the regression line for the relationship at the local level.  However, for 6 relationships, the national measures exhibit the reverse relationship.  The sections below explicitly point out these apparent anomalies wherever they occur.  In all cases, additional analyses separately examined the local level relationship within each study and verified that the local level relationship within each study conformed to the local level relationship in the overall, pooled database (i.e., the slope of the regression lines in the separate studies always followed the same slope as the overall regression line computed from the pooled data).  The apparent anomalies appear to derive from the fact that the two studies regression lines are at markedly different elevations, which in turn probably reflects the influences of other factors.  This is illustrated and discussed further below.

Rate of Alleged Maltreatment

The national rate of alleged maltreatment showed no statistically reliable change from the time of the LAS to the time of the SPM. However, two agency characteristics significantly related to the rate of alleged maltreatment, measured by the number of children alleged to be maltreated on investigated CPS reports per 1,000 children residing in the agency jurisdiction (Table 3-2).  It was very common for CPS agencies to report that their investigation workers review prior CPS records for all cases. Agencies that did so had significantly higher rates of alleged maltreatment in their jurisdictions.

Figure 3-1 depicts the relationship between the rate of alleged maltreatment and whether investigating workers reviewed prior CPS records for all cases.  The positive slope indicates that the rate of alleged maltreatment, plotted on the Y-axis, was higher in the jurisdictions of agencies where investigating workers did so. Thus, the higher the percentage of agencies with this practice (plotted on the X-axis), the higher the rate of alleged maltreatment.

Figure 3-1. Relationship Between the Percentage of Agencies that Reviewed Prior CPS Records for All Cases and the Rate of Alleged Maltreatment

The two study points in the graph, marked by stars, show that even though this practice was very common at the time of both studies, its use increased significantly from 91% of local CPS agencies in the 2002 LAS to 97% in the 2005-2006 SPM. The estimated national rate of alleged maltreatment also increased between the LAS and the SPM, rising from 42.4 to 43.8 children per 1,000.  The distance between the starred points and the regression line reflects both estimation errors[9] and the influences of other factors that affect the rate of alleged maltreatment.

On the other hand, whether the local CPS agency had sole responsibility for investigating child fatalities was negatively related to the rate of alleged maltreatment.  Agencies that did so investigated lower rates of alleged maltreatment in their jurisdictions, so when a higher percentage of agencies had sole investigative responsibility for fatality investigations, the rate of alleged maltreatment was lower. Nationally, the percentage of agencies that did so decreased from 12% to 5% between the 2002 LAS and the 2005-2006 SPM.  At the same time, the estimated national rate of alleged maltreatment during the same period increased, as reported above and shown for the starred point in the figure.  Again, the national measures (starred points) follow the direction of the relationship at the local-jurisdiction level. However, the fact that they are not directly on the local regression line indicates that other factors also account for the overall measure and that there is some degree of estimation error in all components of the graph.

Figure 3-2. Relationship Between the Percentage of Agencies With Sole Responsibility for Investigating Fatalities and the Rate of Alleged Maltreatment

Percentage of Children Assigned a Substantiated or Indicated Disposition

The percentage of children on investigated reports that CPS assigned a substantiated or indicated disposition showed no statistically reliable change from the time of the LAS to the time of the SPM.  However, this measure was related to three agency characteristics, as Table 3-3 shows.  This maltreatment measure was positively associated with the agency having sole responsibility for investigating severe physical abuse and negatively related to the agencys use of a state hotline to handle phone calls, whether during weekday evenings or on weekends.

Agencies that had sole responsibility for investigating severe physical abuse substantiated/indicated higher percentages of the children they investigated, so the regression line in Figure 3-3 shows that when more agencies have this responsibility a higher percentage of investigated children are assigned a substantiated or indicated disposition.  The percentage of agencies that had sole investigation responsibility for severe physical abuse decreased from 23% to 7% between the 2002 LAS and the 2005-2006 SPM, while at the same time there was a fall in the national estimate of the percentage of children in investigations whose maltreatment was substantiated or indicated from 28% to 25%, as the starred points in this figure indicate.

Figure 3-3. Relationship Between the Percentage of Agencies That Had Sole Responsibility for Investigating Severe Physical Abuse and the Percentage of Children in Investigated Reports Who Were Assigned a Substantiated or Indicated Disposition.

Agencies that used a state hotline to handle calls, whether during weekday evenings or on weekends, substantiated or indicated lower percentages of the children in their investigated cases.  Figure 3-4 shows the pattern for hotline screening during weekday evenings; the graph for weekend hotline screening (not given here) is virtually identical. The starred points in this figure show that the percentage of agencies that used this screening approach increased from 18% in the 2002 LAS to 41% in the 2005-2006 SPM.  There was a concomitant decrease in the estimated national percentage of children assigned a substantiated or indicated disposition during the period, from 28% to 25%.

Figure 3-4. Relationship Between the Percentage of Agencies that Used a State Hotline to Handle Calls on Weekday Evenings and the Percentage of Children in Investigated Reports Who Were Assigned a Substantiated or Indicated Disposition.

Rate of Physical Abuse Alone

The rate of physical abuse alone (i.e., with no other type of maltreatment) decreased from the time of the LAS to the time of the SPM, from 1.9 to 1.5 children per 1,000, a statistically marginal decrease. Six agency characteristics were significantly related to this maltreatment rate: 2 characteristics pertain to the agencys intake and screening, 3 concern activities that the agency reported as always being obstacles to the timely and accurate completion of investigations, and 1 was a service the agency offered after investigation regardless of the investigation outcome.  Table 3-4 identifies these characteristics and gives their regression parameters, indicating the significance of their relationships to the rate of physical abuse alone.

Agencies that had a state hotline handle their referrals in off-hours, whether during weekday evenings or on weekends, had lower rates of physical abuse alone in their jurisdictions, measured by the number of children per 1,000 residing in the county that CPS substantiated or indicated with physical abuse alone (and no other type of maltreatment).  Lower rates of physical abuse alone also occurred in jurisdictions of agencies that reported certain activities were obstacles to their completing timely investigations: explaining the consequences of parents actions to them, preparing the case record, and preparing the court record.

Figure 3-5 graphs the strong negative statistical relationship between using a state hotline to handle calls on weekends and the rate of physical abuse alone.  The starred points show that the percentage of agencies using a state hotline to handle calls during weekends increased from the 2002 LAS to the 2005-2006 SPM from 17% to 41%, respectively.  At the same time, the estimated national rate of physical abuse alone decreased from 1.9 to 1.5 children per 1,000.  The pattern depicted in this figure held true, to slightly different degrees, for the relationship between the rate of physical abuse alone and the characteristics listed in the second through third rows in Table 3-4.

Figure 3-5. Relationship between the Percentage of Agencies That Used a State Hotline to Handle Calls on Weekends and the Rate of Physical Abuse Alone.

In contrast, whether or not an agency provided homemaker/chore services after an investigation was positively related to the rate of physical abuse alone (as shown in the last row of Table 3-4), meaning that agencies that provided these services had higher rates of physical abuse alone in their jurisdictions.  Figure 3-6 depicts this relationship.  The starred points indicate that the percentage of agencies providing this service decreased from the time of the 2002 LAS to the 2005-2006 SPM (from 73% to 39%), while at the same time the overall national rate of physical abuse alone decreased from 1.9 to 1.5 per 1,000 children.

Figure 3-6. Relationship Between the Percentage of Agencies That Provided Homemaker/Chore Services and the Rate of Physical Abuse Alone.

Rate of Neglect Alone

The rate of neglect alone, measured by the number of children per 1,000 residing in the agencys jurisdiction that CPS substantiated or indicated with neglect alone, showed no statistically reliable change from the time of the LAS to the time of the SPM.  However, five agency characteristics were significantly associated with this maltreatment rate (Table 3-5). The presence of two characteristics was associated with lower rates of neglect alone: providing an alternative response and when having insufficient time to make a good determination always constituted an obstacle to completing timely and accurate investigations. In contrast, three other agency characteristics were associated with higher rates of neglect alone: requiring investigative workers to make a determination about maltreatment for all children in the family, always reviewing prior CPS records during the investigation process, and always interviewing professionals known to the family.

Agencies that provided alternative response had significantly lower rates of substantiated/indicated neglect alone in their jurisdictions.  Figure 3-7 depicts this statistically negative relationship.

Figure 3-7. Relationship between the Percentage of Agencies That Provided Alternative Response and the Rate of Neglect Alone.

This figure also shows that the percentage of agencies that provided alternative response decreased from 69% to 39% between the 2002 LAS and the 2005-2006 SPM, and during the same time interval the NCANDS data show the rate of neglect alone shifted from 6.0 to 6.3 children per 1,000.  Although, as mentioned above, this is not a statistically reliable difference in the maltreatment measure, the national pattern conforms to the statistically significant relationship found at the level of the local jurisdictions.

As described above, the analyses encountered a few instances where the overall national estimates were opposite what one would expect from the results of the analyses with the pooled database.  This is the case for the relationship between the rate of neglect alone and whether having insufficient time to make a good determination was always an obstacle to timely and accurate investigations.  Table 3-5 shows that this agency characteristic was negatively related to the rate of substantiated/indicated neglect at the local level.  That is, agencies that reported this factor was always an obstacle had lower neglect rates in their jurisdictions, as indexed by the number of children per 1,000 who were substantiated or indicated as victims of neglect alone.  However, in this case, the national findings on this characteristic for each study do not reflect this underlying negative relationship at the jurisdiction level. The national estimates appear to be in the positive direction:  the percentage of agencies that reported insufficient time to always be an obstacle increased from the LAS to the SPM (from 12% to 34%) at the same time the estimated rate of neglect alone also increased (from 6.0 to 6.3 children per 1,000).

As promised above, further analyses verified that the negative relationship held true within each component study, and that the anomaly derived from the fact that the two studies regression lines are at considerably different elevations.  Figure 3-8 illustrates this, showing that the two studies regression lines both document the negative relationship.  This figure also shows how the national estimates reflect the different elevations of the regression lines in the two studies, with the overall rate of neglect notably higher in the SPM than in the LAS.  This may reflect the fact that that the maltreatment measure is affected by additional factors that these analyses have ignored.  The differential elevation may also stem, in part, from measurement deficiencies due to the missing NCANDS data (which the nonresponse adjustments applied in these analyses could not address).

Figure 3-8. Relationship Between the Percentage of Agencies Where Having Insufficient Time to Make a Good Determination Was Always An Obstacle to Timely and Accurate Investigations and the Rate of Neglect Alone.

Agencies where investigating workers always interviewed professionals known to the family (such as teachers, physicians, clergy) substantiated neglect alone at significantly higher rates, as Figure 3-9 illustrates.  The starred points in this figure show that percentage of agencies that reported this as a consistent activity during investigation increased from 41% in the LAS to 64% in the SPM. The fact that the stars are relatively close to the regression line in this figure shows that the national pattern on this measure tracks the jurisdiction-level relationship relatively well.

Figure 3-9. Relationship between the Percentage of Agencies Where Investigating Workers Always Interviewed Professionals Known to the Family and the Rate of Neglect Alone.

Rate of Sexual Abuse Alone

The rate of sexual abuse alone, which reflects the number of children per 1,000 residing in the county that CPS substantiated or indicated with sexual abuse alone, did not statistically change from the time of the LAS to the time of the SPM.  However, 4 agency characteristics, listed in Table 3-6, significantly related to this maltreatment rate at the level of the local agency jurisdiction.

Agencies that used a state hotline to handle their referrals during non-business hours substantiated or indicated sexual abuse at significantly lower rates in their jurisdictions. This relationship appears in Figure 3-10.  The starred points indicate that the national estimates follow the direction of the jurisdiction-level relationship.

Figure 3-10. Relationship Between the Percentage of Agencies Using a State Hotline to Take Referrals on Weekends and the Rate of Sexual Abuse Alone.

Agencies where investigative workers always reviewed prior CPS records and agencies that provided homemaker or chore services regardless of the investigation outcome substantiated/indicated significantly higher rates of sexual abuse in their jurisdictions.  Figure 3-11 graphs this last relationship.  In this graph, the national estimates again conform to the relationship documented at the level of local jurisdictions.

Figure 3-11. Relationship Between the Percentage of Agencies That Provided Homemaker/Chore Services After Investigation and the Rate of Sexual Abuse Alone.

Not graphed here is the relationship listed in Table 3-5 between reviewing prior CPS records and the sexual abuse rate.  The jurisdiction-level relationship resembles the Figure 3-11 pattern, but the national estimates do not follow the local-agency pattern.  As noted above, this type of discrepancy may stem from the fact that the adjustments applied here for missing NCANDS data did not eradicate biases in the national sexual abuse rate estimates.  Also, the discrepancy may mean that other factors have more important influences on the national sexual abuse rate, which these simple binary analyses have ignored.  Again, as with all anomalous findings in this effort, additional analyses within each study verified that both documented a positive relationship at the local level and that the discrepancy at the national level reflected the different elevations of the regression lines in the two studies.

Rate of Psychological Maltreatment Alone

There was no statistical change in the rate of psychological maltreatment from the time of the LAS to the time of the SPM (0.43 to 0.34 children per 1,000, respectively).  However, as Table 3-6 indicates, eight agency characteristics were significantly correlated with the rate of psychological maltreatment alone, measured by the number of children per 1,000 residing in the county that CPS substantiated or indicated with psychological maltreatment alone (not in combination with another form of maltreatment).  Six characteristics were negatively related to the rate of substantiated/indicated psychological maltreatment, meaning that agencies with the characteristic in question substantiated/indicated lower rates of psychological maltreatment in their jurisdictions.  In contrast, two characteristics were positively correlated with the psychological maltreatment rate, so agencies with the characteristic substantiated/indicated significantly higher rates in their jurisdictions.

Agencies that routed calls or referrals directly to another agency during weekday evenings and on weekends substantiated/indicated significantly lower rates of psychological maltreatment. Similarly, rates of psychological maltreatment were significantly lower in jurisdictions of agencies that had sole responsibility for investigating truancy, those that always conducted criminal background checks during investigation, and those that reported they always encountered obstacles in determining what happened to the child and in spending sufficient time with the family during investigations.  Figure 3-12 illustrates the pattern for this last relationship.

Figure 3-12. Relationship Between the Percentage of Agencies Reporting that Insufficient Time with the Family Was Always an Obstacle to Completing Investigations and the Rate of Psychological Maltreatment Alone.

The starred points in this figure show that the percentage of agencies that reported insufficient time with the family to always be a problem in completing investigations decreased significant from 20% in the 2002 LAS to 9% in the 2005-2006 SPM. At the same time, the estimated national rate of psychological maltreatment at the time of these studies increased from 0.34 to 0.43 children per 1,000, a pattern that generally conforms to the direction of the relationship among local jurisdictions.

Similar graphs describe the other agency characteristics that are negatively related to the rate of psychological maltreatment alone, with the exception of the use of other agencies to handle referrals during non-business hours.  As regards the use of other agencies to handle calls during evenings and weekends, the national estimates from the two studies do not follow the negative relationship found at the local-agency level. Additional analyses verified that both studies indicated a negative relationship at the local level and documented very different elevations of the regression lines in the two studies.  As discussed above, this apparent anomaly may stem from the fact that the adjustments applied here for missing NCANDS data did not eradicate biases in the national sexual abuse rate estimates. Also, this type of discrepancy may mean that other factors have more important influences on the psychological maltreatment rate, which the simple binary analyses reported here have ignored.

Two agency characteristics were positively related to the rate of psychological maltreatment:  whether the agency had an excessive investigation workload and whether it provided the service of a dental exam after an investigation, regardless of the investigation outcome.  Figure 3-13 graphs the relationship between having an excessive investigation workload and the rate of substantiated/indicated psychological maltreatment.  The starred points in this figure show that the national measures generally conform to the pattern found among local jurisdictions on these measures:  the percentage of agencies that reported excessive investigation workloads decreased significantly from 64% at the time of the LAS to 47% during the SPM, while at the same time the estimated national rate of psychological maltreatment also decreased.

Figure 3-13. Relationship Between the Percentage of Agencies Reporting an Excessive Investigation Workload and the Rate of Psychological Maltreatment Alone.

Rate of Medical Neglect Alone

From the LAS in 2002 to the SPM in 2005-2006, there was a statistically marginal increase in the rate of medical neglect, measured by the number of children per 1,000 residing in the agencys jurisdiction that CPS substantiated or indicated with medical neglect alone (from 0.14 to 0.25 children per 1,000, respectively).  At the local jurisdiction level, the 8 agency characteristics listed in Table 3-8 were significantly related to this rate.

Agencies that provided alternative response substantiated significantly lower rates of medical neglect in their jurisdictions, and the national pattern reflected the local jurisdiction relationship.  As Chapter 2 reported, significantly fewer agencies in the nation provided alternative response at the time of the SPM compared to the time of the earlier LAS.  As noted above, the rate of substantiated or indicated medical neglect nationwide increased during this interval.

Two investigative practices related positively to the medical neglect rate: whether the agency required investigative workers to make a determination about maltreatment for all children in the family, and whether it was required that they refer the family for further services if needed. Agencies that required these practices had significantly higher rates of substantiated/indicated medical neglect in their jurisdictions.

Five investigative activities related to the rate of medical neglect, all in the positive direction:  whether, for all cases, investigators reviewed prior CPS records, interviewed family members other than the caregiver, interviewed professionals known to the family, conducted criminal background checks on alleged perpetrators, and discussed the case with other CPS workers.  In jurisdictions where these activities were standard in all investigations, rates of substantiated/indicated medical neglect were significantly higher.  Figure 3-14 illustrates this last relationship.

Figure 3-14. Relationship Between the Percentage of Agencies Where Investigators Discussed All Cases With Other CPS Workers and the Rate of Medical Neglect Alone.

Rate of Multiple Maltreatment

There was no statistically reliable change in the national estimate of the rate of children per 1,000 whom CPS substantiated or indicated with multiple types of maltreatment (1.41 in the LAS vs. 1.54 in the SPM).  However, Table 3-9 shows that 6 agency characteristics significantly related to this maltreatment rate at the local agency level.

Agencies that provided alternative response and where spending sufficient time with the family was always an obstacle to completing investigations substantiated/indicated significantly lower rates of multiple maltreatment in their jurisdictions.  Figure 3-15 shows the pattern for providing an alternative response.  The starred points indicate the national estimates conform to the direction of the relationship at the local jurisdiction level.

However, the national estimates regarding the percentage of agencies that reported insufficient time with the family to always be an obstacle to completing investigations do not conform to the pattern at the local level for that measure (not graphed here).  As noted above, this type of discrepancy may derive from the fact that the adjustments applied here for missing NCANDS data could not adequately correct the national multiple maltreatment estimates and/or other factors may mitigate or modify this relationship at the national level.  Again, additional analyses confirmed that the local-level relationship was negative in both studies, with the national estimates modified by the different elevations of the separate regression lines.

Figure 3-15. Relationship Between the Percentage of Agencies That Provided an Alternative Response and the Rate of Multiple Maltreatment.

Agencies where a state hotline handled referrals during non-business hours, where investigative workers reviewed prior CPS records for all cases, and where investigation workloads were excessive had significantly higher rates of substantiated/indicated multiple maltreatment in their jurisdictions.  Figure 3-16 depicts this relationship for the investigation activity of reviewing prior CPS records.

Figure 3-16. Relationship Between the Percentage of Agencies Where  Investigating Workers Reviewed Prior CPS Records for All Cases and the Rate of Multiple Maltreatment.

The graph shows that the national estimates follow the local-agency pattern of the relationship between reviewing prior CPS records for all cases and multiple maltreatment rates.  However, the national estimates on the percentage of agencies reporting excessive investigation workloads and the rates of multiple maltreatment in the LAS and SPM (not graphed here) do not conform to the local-level pattern, so missing data adjustments and effects of other factors may modify this observed relationship in the two studies at the national level.  As before, further analyses confirmed that the relationship was positive in both studies, with the different elevations of the maltreatment measure in the two studies displaced the national estimates.

Conclusion

This study identified a number of noteworthy changes in local CPS agency structures and practices between the time of the 2002 LAS and the 2005-2006 SPM and uncovered a number of significant relationships between agency characteristics and measures of maltreatment in their jurisdictions.

Changes in local CPS agency structures and practices.  Numerous noteworthy changes occurred in local CPS structures and practices over the short time interval between 2002 and 2005-2006. Among the most notable findings was that the percentage of local agencies that said they provided an alternative (non-investigative) response decreased precipitously from the LAS to the SPM. At the same time, those agencies that provided alternative response at the time of the later study indicated more standardized practices during their response in that they were proportionally more likely to report that this response included assessments and service referrals and that they always conducted specific activities during this response, such as interviewing caregivers and children, visiting the family, and discussing the case with other workers.  They were also less likely to have specialized staff dedicated to their alternative response function and were more likely to indicate that a number of factors posed obstacles to the timely completion of their alternative response option.

Also, during the 2002 to 2006 period, it appears that the screening/intake function became more centralized, with fewer local CPS agencies conducting their own screening/intake and more relying on state hotlines for screening and intake, especially during non-business hours.  At the time of the later SPM, agencies were proportionally more likely to report that they most commonly received referrals from law enforcement agencies and schools but less likely to identify individuals as their most common referral source. Also, during the later timeframe, agencies were proportionally less likely to use non-English speakers on staff to handle calls from non-English speakers but more likely to handle such calls using multiple methods.

Similarly, there was greater standardization of investigation practices, with proportionally more agencies during the later period indicating that investigations always included certain practices and that their workers always conducted specific activities during an investigation. For instance, a higher percentage of agencies during the later period said they always assessed whether one or more children were at risk of maltreatment and determined if all children in the family had been maltreated. At conclusion of an investigation, agencies in the SPM were proportionally more likely to inform the person who reported the alleged maltreatment and were more likely to provide short-term services including parenting classes and substance abuse programs. Significantly higher percentages of agencies indicated that a number of factors posed obstacles to their timely and accurate completion of investigations during the 2005-2006 timeframe.  At the same time, more agencies offered parenting classing and substance abuse programs after an investigation, but fewer offered other specialized services, such as marital counseling, family system therapy, grief counseling, advocacy services, dental exam, homemaker/chore programs, financial planning, or employment services.

With regard to collaboration with other agencies, agencies were proportionally less likely in 2005-2006 to have sole responsibility for investigating more serious forms of maltreatment, since more shared such responsibility with law enforcement or other agencies.  At the same time, more agencies had sole responsibility for investigating truancy but shared responsibility with non-law enforcement agencies for investigating maltreatment perpetrated by a foster parent.

Relationships between agency characteristics and maltreatment.  A number of agency characteristics that changed significantly between the 2002 LAS and the 2005-2006 SPM were related to rates of child maltreatment in the local agencies jurisdictions. Providing an alternative response function, using a state hotline to handle referrals during nonbusiness hours, and consistently reviewing prior CPS records during investigations were characteristics that demonstrated pervasive relationships to maltreatment rates across a number of measures.  Agencies that provided an alternative response had significantly lower rates of neglect, medical neglect, and multiple maltreatment in their jurisdictions.  Agencies that always reviewed prior CPS records during investigations had higher rates of maltreatment in their jurisdictions on a number of measures, including the overall rate of alleged maltreatment on reports to CPS and rates of sexual abuse, neglect, medical neglect, and multiple maltreatment. Agencies that used a state hotline to handle calls during non-business hours substantiated/indicated a significantly lower percentage of children in their investigated reports and had significantly lower rates of substantiated/indicated physical abuse and sexual abuse in their jurisdictions.  However, these agencies also had significantly higher rates of substantiated/indicated multiple maltreatment.

Achievements, limitations, and future possibilities.  This research is the first study to empirically examine changes in local CPS agency structure and practice using nationally representative samples. Despite the fact that the time interval between assessments was relatively short (3-4 years), it identified a number of changes in agency operations. Further, it uncovered a number of relationships between local agency characteristics and measures of child maltreatment in local agency jurisdictions. These findings here are solely descriptive.  However, they may suggest how, if a given relationship and other influencing factors remain stable, substantial and significant changes in the prevalence of certain agency characteristics or practices may relate to changes in maltreatment measures.

The findings on the relationship between agency characteristics and maltreatment measures are qualified by several limitations.  First, NCANDS data were not available for a number of local agencies  with more data missing for agencies in the earlier LAS sample than for agencies assessed during the later SPM timeframe.  Although attempts were made to adjust for these missing data, such adjustments are always approximations and missing data always has the potential to introduce biases.  How such biases may have affected the current study is not known, but this consideration certainly qualifies the reported findings.  Second, only those agency characteristics that showed significant changes between 2002 and 2006 were explored in relation to maltreatment measures.  Other agency characteristics that changed little or remained stable over the time interval may have equally strong or even stronger associations to rates of maltreatment in local jurisdictions.  Third, no multivariate models were computed for this report. However, agency characteristics probably cluster together and maltreatment rates in local jurisdictions are undoubtedly related to many factors simultaneously.

Further analyses of these data could explore how agency characteristics cluster together and how multiple agency characteristics simultaneously relate to local maltreatment rates.  For example, does the fact that fewer agencies reported less concern about excessive workloads in investigation during the later time period relate (1) to the fact that more of them also report sharing investigation responsibility with law enforcement or other agencies, (2) to their greater standardization of investigation activities (i.e., the fact that more agencies always follow certain investigation practices), or (3) to the fact that fewer agencies report offering specialized services after an investigation?  Alternatively, does the fact that more agencies identify circumstances that pose obstacles to timely and accurate completion of investigations relate to the fact that more activities are standard practice during investigations?  Additional analyses across both studies data could address questions such as these by examining the relationships among these agency characteristics at the local level.  Additional regression analyses could also examine whether and how agency characteristics that did not change over the LAS-SPM interval relate to maltreatment rates.

[ Go to Contents ]

References

Li, S., Shusterman, G., and Sedlak, A.  (2009).  Fourth National Incidence Study of Child Abuse and Neglect (NIS-4):  CPS Structure and Practices Mail Survey (SPM).  Rockville, MD:  Westat, Inc.

U.S. Department of Health and Human Services. (2003). National Study of Child Protective Services Systems and Reform Efforts: Findings on Local CPS Practices. Washington, D.C. U.S. Government Printing Office. Online at: http://aspe.hhs.gov/hsp/CPS-status03/CPS-practices03/index.htm>

[ Go to Contents ]

Endnotes

[1] The LAS also asked about key changes in administrative structure and practices related to various reform efforts at the time of the survey. To reduce the burden of response for local CPS agencies, the SPM did not ask these questions. The comparison analyses reported here are limited to the items that are common to the two surveys.

[2] The results for the LAS in this table depart markedly from those in Table 2-12 in the original LAS report.  This is because the analysis here computed the percentage for each cell in this table independently, using all of the agencies that conducted the function in question. Thus, the sample sizes varied by both function and study, depending on number of agencies that provided the function in that survey. In contrast, the analysis for the LAS report (Table 2-12 there) was restricted to a small subset of the sample (i.e., agencies that reported workload concerns only regarding a single function).

[3] The estimated percentages of agencies in the Other methods categories of this table are much larger than the percentages in the same categories of Table 3-3 of the LAS report. The discrepancies are due to a difference in coding the responses of Other (specify). In this report, most of those responses were assigned to the Other methods category.

[4] For those agencies that served only part of a county, the county population and the NCANDS cases in the county were allocated to the agency as detailed in Appendix C.

[5] Preliminary work also attempted to include three additional measures from NCANDS data: the percentage of children alleged to maltreated that CPS investigated, the number of children who received an alternative response and the number who received both an investigation and an alternative response. However the first of these measures was omitted from the final analyses because most counties investigated all, or nearly all, children in investigated reports, so there was not sufficient variation on this measure to reveal informative patterns.  The last two measures were not used in the final analyses because the majority of the LAS and SPM counties were missing values on these measures for the study years.

[6] These analyses excluded agency practices that reflected different approaches to implementing alternative response.  Agencies that do not conduct alternative response at all must be excluded from analyses of such measures.  Excluding those agencies would have eliminated the majority of agency records in the database.  Findings based on such extremely censored data are too unreliable to report.

[7] Chapter 2 reported a significant difference between the percentages of agencies in the LAS and SPM that reported problems locating the family always posed an obstacle to completing timely and accurate investigations.  However, very few agencies reported this to be a problem in either study, and NCANDS maltreatment data were missing for all but 3 of these agencies.  Thus, the analyses reported in this chapter excluded this agency characteristic.

[8] Note that the points do not reflect actual NCANDS national measures for the study years, but because of the adjustments described in Appendix C, the pooled database may generate national estimates that correspond closely to the national findings from NCANDS.

[9] There is a component of error in these estimates because all the information derives from samples and because of missing data, which may still introduce some bias despite efforts to address the lost information.