Assessing the Context of Permanency and Reunification in the Foster Care System. 3.1.2 Reunification During Different Periods in the Duration of a Spell

12/01/2001

Table IIC.2, is a modification of the reunification analysis from Table IIC.1, designed to investigate the influence of duration in care and the timing of exits on the relationship between the independent variables and reunification. This design is motivated by the observations made earlier (from the IIB tables) that reunification levels are much higher during the early months of spells in foster care. This model examines whether the factors related to reunification also change as the child remains in care over time.

  • The first three models consider the likelihood of reunification for all spells that started in 1990-94. Column I is a model of the likelihood of reunification within the first six months of the spell. Column 2 models the likelihood of reunification within the first 18 months of the spell, and Column 3 the likelihood of reunification within 3 years. (This column repeats the reunification results from the first model inTable IIC.1).
  • The next two models consider the likelihood of reunification only for those spells that remain open for at least six months, i.e., the child has not exited from care by reunification or any other type of spell discharge. Column 4 is a model of the likelihood of reunification between 6 and 18 months. Column 5 a model of the likelihood of reunification between 6 months and 3 years.
  • The final model considers the likelihood of reunification by 3 years only for those spells that remain open for 18 months or more.

Overall, the basic structure of the odds ratios is rather similar in all six of these models, suggesting that the influence of these predictor variables on reunification is somewhat stable for all portions of a spell in foster care. However, some differences can be observed.

Table IIC.2
Logistic Regression Model of Reunification as Outcome of Substitute Care Placement Episodes. Selected Time Periods. Pooled Individual Spell Data From 9 States, Entry Cohorts 1990-1994.
Predictor variable Category From Time of Spell Entry: If 6 months elapsed without exit: If 18 months without exit:
Reunification before 6 months Reunification before 18 months Reunification before 36 months Reunification before 18 months Reunification before 36 months Reunification before 36 months
Odds Ratios Odds Ratios Odds Ratios Odds Ratios Odds Ratios Odds Ratios
Age at Entry <3 months 0.48 0.43 0.39 0.46 0.41 0.41
3-11 months 0.80 0.82 0.78 0.88 0.80 0.76
1-2 years 0.92 0.96 0.94 1.02 0.97 0.92
3-5 years 0.95 0.99 0.98 1.03 1.00 0.97
6 - 8 years 0.95 0.99 0.98 1.03 1.00 0.97
9-11 years * 1.00 1.00 1.00 1.00 1.00 1.00
12-14 years 0.88 0.90 0.83 1.06 0.96 1.02
15-17 years 0.70 0.69 0.57 0.95 0.70 0.70

Gender

Male 0.92 0.99 1.01 1.03 1.04 1.05
Female * 1.00 1.00 1.00 1.00 1.00 1.00

Race/Ethnic

African American 0.69 0.63 0.65 0.62 0.66 0.77
Hispanic 1.12 1.02 1.00 0.93 0.93 0.97
White and  Other * 1.00 1.00 1.00 1.00 1.00 1.00

Urban

Primary urban county 0.70 0.62 0.65 0.58 0.64 0.74
Remainder of State * 1.00 1.00 1.00 1.00 1.00 1.00

Sequence

First Spell * 1.00 1.00 1.00 1.00 1.00 1.00
Reentry Spell 0.67 0.72 0.72 0.90 0.87 0.93

Care Type

Congregate Care 1.07 1.18 1.18 1.47 1.50 1.65
All others  (FC, KC)* 1.00 1.00 1.00 1.00 1.00 1.00

Stability

One placement in spell * 1.00 1.00 1.00 1.00 1.00 1.00
2+ placements 0.24 0.33 0.40 0.43 0.48 0.56

State:

Alabama 0.82 0.45 0.36 0.33 0.26 0.26
California 1.24 1.22 1.19 0.88 0.84 0.66
Illinois 0.88 0.71 0.68 0.51 0.50 0.45
Maryland 1.21 0.80 0.72 0.46 0.46 0.47
Michigan * 1.00 1.00 1.00 1.00 1.00 1.00
Missouri 1.93 1.29 1.17 0.82 0.83 0.85
New Mexico 2.32 1.63 1.37 1.12 0.96 0.73
New York 0.95 0.88 0.96 0.74 0.87 0.97
Wisconsin 1.44 1.14 1.45 1.20 1.26 1.15
Year of Entry 1990 1.14 1.12 1.12 1.06 1.08 1.07
1991 1.10 1.06 1.05 1.01 1.02 1.00
1992 * 1.00 1.00 1.00 1.00 1.00 1.00
1993 1.02 0.99 0.98 0.96 0.96 0.95
1994 0.94 0.94 0.92 0.94 0.91 0.87
  Intercept -0.34 0.49 0.78 -0.14 0.42 -0.57
exp(intercept) 0.71 1.63 2.18 0.87 1.52 0.57
Predictive Concordance 0.73 0.72 0.70 0.70 0.69 0.66
N spells 404,416 404,416 404,416 272,725 272,725 187,286
N successes 89,242 147,085 176,353 57,843 87,111 29,268
Proportion successes 0.22 0.36 0.44 0.21 0.32 0.16

Note: Predictors noted with "*" are not contained in model, but are the "excluded" category for their variable. Predetermined odds ratios of 1.00 are assigned to these categories. Odds ratios for other categories of the same variable express effects in relation to that of the excluded category.


For example, the racial gap in the odds of reunification narrows markedly as spell duration increases. Between 0 and 6 months, Hispanic children are 62 percent more likely than African American children to be reunified, and 12 percent more likely than white children. Between 6 and 18 months, Hispanics are 50 percent more likely than African Americans, and 7 percent less likely than whites to be reunified. Between 18 months and 3 years, Hispanics are 26 percent more likely than African Americans and 3 percent less likely than whites to be reunified.

Similarly, children in reentry spells and children in spells with multiple placements are characterized by very low relative likelihoods of reunification during the first six months in care. In each case, as the spell duration increases, the odds of reunification for these categories grow closer to (but not equal to) those of children in first spells and children in single-placement spells.

The largest changes appear in the odds ratios for states. New Mexico and Missouri both have highest relative odds of reunification within the first six months of a spell, yet moderate to low relative odds of reunification between 18 months and 3 years. Thus, early reunification appears to be unusually likely in Missouri. Michigan, in contrast has moderately low relative odds of reunification between months 0 and 6, but the second highest odds of reunification between months 6-18 and 18-36.

Overall, the influence of the substantive covariates of reunification appear to be relatively stable for different temporal portions of foster care spells, and when they do change, they tend to become less influential as the time in care increases. Thus early reunifications appear to be more readily predicted by child characteristics and case factors than do late reunifications. Also, we see that the states vary widely in the likelihood of reunification for spells of different lengths. This is similar to the wide variation in the timing of reunification that was observed in Figure C.