Assessing the Context of Permanency and Reunification in the Foster Care System. 1.4 Summary

12/01/2001

Table IIA.1presented nine sets of two-way tables. Taken together, they demonstrate that there is systematic and substantial variability in the patterns of exit from foster care spells. Of all nine variables, only gender did not produce an important qualification about the use of reunification to end spells in foster care. Variables with particularly strong relationships were state, age at entry, and spell duration. Some of the more important relationships are portrayed graphically in Figure B.

Figure B 
Proportion of Observations Exit from Foster Care, By Destination and Selected Characteristics 1990-1994. Entries Observed Through December 31, 1997. Nine States.

Exit Type by Age at Entry 

Figure B1.  Proportion of Observations Exit from Foster Care, By Destination and Selected Characteristics 1990-1994. Entries Observed Through December 31, 1997. Nine States.

Exit Type by Placement Type 

Figure B2.  Proportion of Observations Exit from Foster Care, By Destination and Selected Characteristics 1990-1994. Entries Observed Through December 31, 1997. Nine States.

Exit Type by Race/Ethnicity 

Figure B3.  Proportion of Observations Exit from Foster Care, By Destination and Selected Characteristics 1990-1994. Entries Observed Through December 31, 1997. Nine States.

Exit Type by Spell Duration 

Figure B4.  Proportion of Observations Exit from Foster Care, By Destination and Selected Characteristics 1990-1994. Entries Observed Through December 31, 1997. Nine States.

As observed by their relationships with the other variables, it appears that reunification exits and family exits are related -- in that they tend to occur commonly to the same type of children and in similar case situations. These observations suggest that we might consider them together as permanency solutions. This will be examined in a following section where they are combined into a single category termed "family exits."

However, these bivariate findings also pose questions. One issue is the high interrelation among the independent variables (multicollinearity). For example, placement stability is highly correlated with duration and with placement type. The relationships seen in the two-way tables between exit type and any of these variables might easily be caused by one of the other related predictors, by a combination of the predictors, or by another related factor that is not described. To address this issue of multicollinearity, multivariate analyses will be presented later in the report that use the predictor variables simultaneously.

Another issue is interpreting what a variable actually represents. For example, age at entry can capture many dimensions of a foster care case. The high level for adoption of infants is partially due to child characteristics (e.g., their attractiveness on the adoption market). But it is also true that the family situations that lead to the removal of infants also involve a large share of cases where terminating parental rights and initiating the adoptive process is thought to be necessary. These tables demonstrate that reunification is complex and multidimensional, and we see that interpretations should be made with care.