Establishing Case Typologies in Foster Care Susan J. Wells, Margaret Skrypek & Kimberly Ford A project conducted by the Gamble-Skogmo Chair in Child Welfare and Youth Policy University of Minnesota School of Social Work Susan J. Wells, PI; Research Assistants: Margaret Griesgraber, Kimberly Ford, Tamara Kincaid, Lucy Alderton, Jose Diaz, Kristin Bauerkemper, Laura Koranda, Amber Link Funded by the USDHHS Children’s Bureau, Grant # 90-CA-1719/02
Need for the Study: Emphasis on Outcomes in Child Welfare Helps us think more systematically about intervention and measuring progress of children and families and the systems that serve them Begins to lay the basis for identifying and documenting evidence-based practice – however…. Can cause confusion in comparisons among local and state child welfare agencies January 20072Contact: Susan J. Wells University of Minnesota School of Social Work
The Value of Case Typologies Help workers apply appropriate intervention strategies Allow agencies to measure success by identifying the probable outcomes of various case types Permit workers to understand changing family and community trends and anticipate service needs Facilitate tracking and comparing data across programs January 20073Contact: Susan J. Wells University of Minnesota School of Social Work
Historically, Major Classifications in Developing Case Typologies Related to Child Welfare…. 1. According to the characteristics of the parent(s) 2.According to environmental factors 3. Based on child characteristics January 20074Contact: Susan J. Wells University of Minnesota School of Social Work
Research Question Is it possible to identify different types of cases in foster care that are reliably associated with different outcomes? If yes, what are these types and how are outcomes affected? January 20075Contact: Susan J. Wells University of Minnesota School of Social Work
Overview of Methods State administrative survey AFCARS analysis Worker/Supervisor focus groups January 20076Contact: Susan J. Wells University of Minnesota School of Social Work
Child Welfare Worker Perspectives A qualitative study of focus group data gathered in five counties in Minnesota January 20077Contact: Susan J. Wells University of Minnesota School of Social Work
Demographics of the Focus Groups Five Minnesota counties Focus groups of 4 to 12 participants Mean number of participants = 8.6 Mean participation rate = 10.5% Majority of participants were Caucasian women Child welfare experience of participants ranged from 2 months to 25 years Mean years of experience = 7.6 yrs January 20078Contact: Susan J. Wells University of Minnesota School of Social Work
Data Collection Focus Group participants were asked to identify: types of cases they encounter other than abuse and neglect cases characteristics of these cases services offered to clients in these cases outcomes of the cases Participants were asked to be specific and to respond in their own words. January 20079Contact: Susan J. Wells University of Minnesota School of Social Work
Data Analysis Coding process Each researcher individually coded the data from the first focus group Researchers compared categories to check for interrater reliability Each researcher coded data for the two focus groups she attended Researchers used data from individual counties to create a comprehensive typology Researchers conducted second-level coding January Contact: Susan J. Wells University of Minnesota School of Social Work
Study Findings Final Typology of 18 case types within four broad categories: Parent’s Capacity or Behavior Child Reasons Problems in Parenting Previous Unsuccessful Child Welfare Intervention January Contact: Susan J. Wells University of Minnesota School of Social Work
Case Types by Category Parent’s capacity or behavior Parent’s inability to care for child because of parent’s mental health Parent’s inability to care for child because of parent’s medical reason/disability Parent’s inability to care for child because of parent’s substance abuse Child is exposed to domestic violence Parent absent from the home January Contact: Susan J. Wells University of Minnesota School of Social Work
Case Types by Category (cont.) Child Reasons Parent’s inability to provide care because of child’s medical reasons Parent’s inability to care for child because of child’s mental health Child behavior (including child substance abuse) January Contact: Susan J. Wells University of Minnesota School of Social Work
Case Types by Category (cont.) Problems in Parenting Physical abuse: parent or caretaker Sexual abuse: parent or caretaker Emotional abuse Parent abandons child Neglect (unsafe in home, basic needs unmet, etc.) Physical or sexual abuse (parent’s failure to protect from perpetrator) Parent–child conflict January Contact: Susan J. Wells University of Minnesota School of Social Work
Case Types by Category (cont.) Previous unsuccessful child welfare intervention Pregnant mother with prior involuntary termination of parental rights Failed adoptive placements Relative lack of follow-through in transfer of custody January Contact: Susan J. Wells University of Minnesota School of Social Work
Factors Affecting the Level or Type of Service Age of child Vulnerability of the child Court involvement January Contact: Susan J. Wells University of Minnesota School of Social Work
How Outcomes Differ by Case Type Higher rates of re-entry: Cases involving parents with chemical dependency Cases that are particularly complex Participant responses were inconsistent regarding cases with lowest rates of re-entry January Contact: Susan J. Wells University of Minnesota School of Social Work
Advantages and Limitations of Focus Group Study Case typology used child welfare workers’ perspectives Rich level of data with important detail regarding case types Qualitative methodology with no objective test of validity Sample population from MN only - not generalizable January Contact: Susan J. Wells University of Minnesota School of Social Work
AFCARS Analysis Analyzed AFCARS data from 10 states using two- step cluster analysis to determine if certain types of cases cluster together Using cluster analysis and literature on different case types, identified groups of case characteristics to develop 19 case types Combined the case types in logistic regressions for each of the 10 states to predict 2 federal outcome measures – placement stability and re-entry into foster care January Contact: Susan J. Wells University of Minnesota School of Social Work
Advantages and Limitations of Using the 2001 AFCARS Data Information on numerous child and case characteristics Cross-sectional snapshot of children in the foster care system rather than change over time Inconsistencies in foster care reporting among states Variation in the demographic composition of the children and in types of cases captured. Quality and consistency of descriptive data collected January Contact: Susan J. Wells University of Minnesota School of Social Work
Cluster Analysis For state sampling, identified homogenous groupings of states based on 3 common variables related to: Demographics Performance Size of foster care system 58 AFCARS variables were used to perform a cluster analysis on data from the selected states Discriminant analysis was used to validate case cluster membership January Contact: Susan J. Wells University of Minnesota School of Social Work
Resulted in over 100 clusters or sub-clusters; Identified variables characterizing clusters… Significant in formation of clusters Representing a large portion of cases in a specific cluster or… Equally distributed among clusters and constituting at least 33% of a cluster Developed maps of relationships of variables in different clusters January Contact: Susan J. Wells University of Minnesota School of Social Work
Final Case Types Were Established by Using the variable/cluster mapping to identify case groupings and then Developing a hierarchy of membership so case types were mutually exclusive Using vars from clusters, beginning with the case characteristics recog. in prior literature as associated with varying case dynamics January Contact: Susan J. Wells University of Minnesota School of Social Work
Percentages of Case Type by State for Children Who Entered Foster Care ILMNPATX Removal reason child disability, White, and all ages 0.0%3.2%2.1%6.1% Removal reason child disability, African American, American Indian or Other Race, and all ages 0.0%0.8%1.6%2.3% Clinical disability, White, and all ages3.2%1.7%1.3%8.8% Clinical disability, African American, American Indian or Other Race, and all ages 5.4%1.0%1.2%4.8% Other medical condition, all race/ethnic groups, and all ages 0.0% January Contact: Susan J. Wells University of Minnesota School of Social Work
Percentages of Case Type by State for Children Who Entered Foster Care (cont.) ILMNPATX Removal reason behavior problems, 11 yrs and older, and all race/ethnic groups 2.2%49.3%39.2%6.0% Removal reason behavior problems, 10 yrs and younger, and all race/ethnic groups 0.4%2.4%2.8%2.4% Emotionally disturbed, all ages, and White or African American 11.2%2.6%3.4%2.0% Emotionally disturbed, all ages, and American Indian or other race 0.8%0.5%0.3%0.0% Removal reason sexual abuse, all ages, and all race/ethnic groups 0.9%2.3%2.7%8.6% Removal reason neglect, 1-10 yrs old, and African American or White 26.0%8.0%9.7%27.4% Removal reason neglect, 1-10 yrs old, and American Indian or other race 2.7%2.0%0.5% January Contact: Susan J. Wells University of Minnesota School of Social Work
Percentages of Case Type by State for Children Who Entered Foster Care (cont.) ILMNPATX Under 1 and all race/ethnic groups18.7%5.0%8.0%9.9% 1-10 yrs old, single female, and African American or White 2.7%5.9%5.3%3.9% 1-10 yrs old, single female, and American Indian or other race/ethnicity 0.3%1.4%0.1% 1-10 years old8.8%4.7%8.9%6.0% 11 yrs old and older, single female, and African American 2.6%1.7% 2.2% 11 yrs old and older, physical abuse, and White, American Indian or other race 2.1%1.3%1.5%1.7% 11 yrs old and older, and all race/ethnic groups 11.8%6.2%9.5%7.6% January Contact: Susan J. Wells University of Minnesota School of Social Work
Odds Ratio Ranking for FC Re-entry (Outcome 4.2) by Case Type (logistic regression) ILMNPAUT RankExp(B)RankExp(B)Rank Exp(B) RankExp(B) Overall [1] : [1] White Child Disability rmved [2] [2] OthrRace Child DisabilityConst [3] [3] not sig. AllRace Child Behavior AllRace Child Behavior 10-not sig White/AA Emot. Disturbed Const3 AI/OthrRace Emot. Disturbed Const2 White Clinical Disability Const AA/AI Clinical Disability Const Other Medical ConditionConst January Contact: Susan J. Wells University of Minnesota School of Social Work
[1] [1] The overall ranking reflects the average ranking when all states (N=9) are taken into account and occurs only for case types that show relative consistency, e.g. individual state ranking is generally within 3 digits across states. [2] [2] Manually removed because the case type contained a low number of frequencies (indicated by an inflated standard error) [3] [3] SPSS automatically removed from model because variable contained 0 frequencies [1] [1] The overall ranking reflects the average ranking when all states (N=9) are taken into account and occurs only for case types that show relative consistency, e.g. individual state ranking is generally within 3 digits across states. [2] [2] Manually removed because the case type contained a low number of frequencies (indicated by an inflated standard error) [3] [3] SPSS automatically removed from model because variable contained 0 frequencies Sexual Abuse not sig. AA/White Neglect AI/Othr Race Neglect not sig. All Races <1Contrast Category AA/Wh Single Female HH not sig. AI/Othr Single Female HH 1-10 not sig rmvd All Races not sig. AA Single Female HH rmvd AI/Wh/Othr Physical Abuse All Race Odds Ratio Ranking for FC Re-entry (Outcome 4.2) by Case Type (logistic regression) ILMNPAUT RankExp(B)RankExp(B)RankExp(B)RankExp(B) Overa ll [1] : [1] January Contact: Susan J. Wells University of Minnesota School of Social Work
Table Notes [1] [1] The overall ranking reflects the average ranking when all states (N=9) are taken into account and occurs only for case types that show relative consistency, e.g. individual state ranking is generally within 3 digits across states. [2] [2] Manually removed because the case type contained a low number of frequencies (indicated by an inflated standard error) [3] [3] SPSS automatically removed from model because variable contained 0 frequencies January Contact: Susan J. Wells University of Minnesota School of Social Work
Some Notable Findings Clusters broke out by age and race of child Clusters highly related to child problems and reasons for entry into care Most likely to re-enter: behavior problem, all races, ages 11 and older Some similarities and some great differences among states in case types and assoc. outcomes January Contact: Susan J. Wells University of Minnesota School of Social Work
Multi-method Integrated Findings Emotional problems, clinically diagnosed disabilities and behavior problems were primary issues in quantitatively identified case typologies and outcomes Yet, the way these cases are served and reported varies greatly among states And fed’l stds apply equally across all states January Contact: Susan J. Wells University of Minnesota School of Social Work
Multi-method Integrated Findings (2) Common multi-method findings Child’s medical, mental health, behavior Sexual abuse Neglect Additional qualitative findings Positive toxicology in newborns (low #s) Failed child welfare intervention (important add.) Child MH and parent/child conflict January Contact: Susan J. Wells University of Minnesota School of Social Work