Policy and Complex Systems
system science to deepen our understanding of human built systems though the unique fine-grained , reliable data that are generated as children move through the foster care system .
Foster care is a resource-constrained system and these resource constraints exist differently within different subsets of foster-care placements . More specifically the management of placements is necessarily different in traditional foster care than it is in congregate care .
Traditional foster care is a placement in the home of a state-licensed substitute caregiver who is often unknown to the child at the time of the placement . The capacity of traditional foster homes represents a resource stock that is managed by the state , requiring active recruitment , licensing , and resource management . Thus , the stock of traditional foster homes is more flexible than those beds available in congregate care . Congregate care , or group care , is placement in a group facility managed by the state or by a private entity contracted by the state . Though congregate care can take several different forms ( e . g ., an in-patient mental health center ), a commonality is that these brick-and-mortar facilities are severely constrained in their ability to grow and shrink the available bed capacity . Thus , in comparison to traditional foster care , congregate care beds are the most inflexible , and , purely in terms of beds , the congregate care system is more resource constrained . If the resource constraints exist and impact flows then some evidence of causal coupling should be found in foster care time series analyzed through population growth models .
This resource-based analysis is described by Wulczyn ( 1996 ), who proposed a methodological framework for analyzing the aggregate level dynamics of foster care populations that sought to define problems related to system change in terms of models capable of directly answering change-based questions . Specifically ,
dx ( t )/ dt = r [ x ×( t ) – x ( t−1 )] ( 1 )
referred to as the linear partial adjustment model , defines the change in population x between times t−1 and t as a variable relationship between the population of x at time t , the rate of population growth r , and “ the level toward which causal forces are impelling [ x ],” defined as the x *( t ) term ( Tuma & Hannan , 1984 , p . 440 ). This model starts with the concept that a system ’ s state at time t−1 impacts that system at time t in a predictable way , and elaborate that to suggest that a portion of that impact and predictability is contained within a target value , x ×( t ). Wulczyn ( 1996 ) hypothesizes that resource utilization serves as target value that both aggregate exits and entries dynamically adjust in accordance to . Recent advances in data availability , computation , and methodology provide an opportunity to test these hypothesized relationships .
Data
The data for this paper were drawn from the Multistate Foster Care Data Archive ( FCDA ), a longitudinal data warehouse maintained by the Center for State Child Welfare Data and Chapin Hall at the University of Chicago . The FCDA contains fine-grained data for individual children and their experiences within the child welfare system , including details about outof-home care spells . A care spell represents the amount of time a child spends in out-ofhome care , and is marked by a start date , or entry , and an end date , or exit . The entry / exit structure of this data set lends itself to the ecological metaphor where birth / death dynamics examine the growth and decline
109