To better understand the family, child, and adoption characteristics associated with payment changes, we drew on the information about adoptive families contained in the California AD 42-R. The sample of cases in which the AD 42R data match with the AAP data includes 771 children under 18 years old. Unlike in North Carolina, most AAP recipients experience periodic payment changes, probably coinciding with recertification.
Exhibit 26 shows basic demographic information on adoptive families. Approximately half of the children were female (53 percent). Over 70 percent of children had been removed from their previous home before they were two years old, and 59 percent had come to live with the adoptive families by the age of two. The majority of both birth mothers (61 percent) and adopting mothers (70 percent) were white not Hispanic. While the percentage of birth mothers with Hispanic origin (23 percent) in the sample was higher than that of African-American birth mothers (14 percent), the percentage of African-American adopting mothers was the same as the percentage of Hispanic adopting mothers (14 percent). In terms of transracial adoption, three-quarters (77 percent) of adopting mothers had the same race as the birth mother. Most of the adopting mothers were high school graduates (29 percent) or had some college or trade school (35 percent). About four-fifths (79 percent) of the adopting parent(s) were not related to the adopted children prior to the adoption. Almost half (48 percent) of the adopting families had two or three minor children, and about half (52 percent) of the adopting mothers were in the third decade of their lives. About half (52 percent) of the adopting mothers were in their thirties, and roughly one-quarter of the mothers (24 percent) were in their forties. Four-fifths (81 percent) of adoptive families had two parents, and just over half (51 percent) of adopting mothers worked outside the home prior to the adoption.
|Characteristics||Sample size (N)||Percentage (%)|
|Gender of child||Male||363||47.1%|
|Age at removal from home||0 - 2||562||74.0%|
|3 - 5||137||18.1%|
|6 or older||60||7.9%|
|Age when child lived with this family||0 - 2||459||59.5%|
|3 - 5||191||24.8%|
|6 or older||121||15.7%|
|Race of birth mother||White||470||61.0%|
|Race of adopting mother||White||543||70.4%|
|Transracial adoption||Same race||594||77.0%|
|Education of adopting mother||Less than high school||104||13.5%|
|High school graduate||223||28.9%|
|Some college/trade school||266||34.5%|
|Four-year-college graduate or more||178||23.1%|
|Relative adoption||Relative adoption||163||21.1%|
|Number of minor children||1||133||17.3%|
|4 or more||269||34.9%|
|Age of adopting mother||< 20||2||.3%|
|60 or more||15||2.0%|
|Single parent||Single parent||148||19.2%|
|Employment of adopting mother||Outside of home||392||51.3%|
Studying monthly payment changes with the AAP database. Before conducting multivariate analyses we further described the sample's involvement with payment changes. We tracked up to five payment changes per case. Exhibit 27 shows how many AAP recipients had experienced each payment change. AAP recipients changed their subsidy levels twice on average with most experiencing one or two payment changes. These data are consistent with the larger population of AAP cases described above, and of which this sample is a subset.
|No payment change||1 payment change||2 payment changes||3 payment changes||4 payment changes||5 payment changes||Mean|
We examined the timing of AAP recipients' payment changes. Change forms without subsidy payment changes, e.g., address changes were excluded from the analysis. Exhibit 28 graphically shows the distribution of monthly payment change amount by the duration of time from the first AAP record. Most AAP recipients have experienced AAP payment changes periodically, about every two years probably coinciding with recertifications. Few of these payment changes were more than $300 per month in either direction.
Exhibit 29 shows average subsidy levels of each payment change by the duration of time since the first AAP record. The average amounts of the first payment change generally decreased except at the fourth year since the first AAP record. On the other hand, the amounts of second payment changes increased up to three years since the first case action and went down again after the fifth year.
|1 year||1 year||2 years||3 years||4 years||5 years||6 years||7 years||8 years||Total|
|1st payment change
(n = 642)
(n = 21)
(n = 89)
(n = 112)
(n = 157)
(n = 81)
(n = 75)
(n = 68)
(n = 32)
(n = 7)
|2nd payment change
(n = 461)
(n = 3)
(n = 20)
(n = 43)
(n = 140)
(n = 80)
(n = 69)
(n = 67)
(n = 32)
(n = 7)
|3rd payment change
(n = 244)
(n = 9)
(n = 20)
(n = 34)
(n = 40)
(n = 54)
(n = 52)
(n = 28)
(n = 7)
|4th payment change
(n = 121)
(n = 3)
(n = 10)
(n = 14)
(n = 23)
(n = 24)
(n = 18)
(n = 24)
(n = 5)
|5th payment change
(n = 66)
(n = 2)
(n = 10)
(n = 9)
(n = 13)
(n = 13)
(n = 12)
(n = 4)
(n = 3)
Bivariate relationships between case characteristics and payment changes. With cross tabulations and chi-square tests, we examined bivariate associations between changes in subsidy level and adoptive families' demographic characteristics (Exhibit 30). This analysis focused on positive amount of payment changes because the negative payment changes were often in response to the positive changes (i.e., they were subsequent corrections or readjustments) and were not independent of them. These analyses examine payment changes as events that signal needs (of varying magnitude) within the adoptive family, rather than focusing on the amount of subsidies received over time. Therefore, the negative amount of each payment change was recoded to zero, and the sum of increased changes from the first payment change to the fifth payment change was considered as positive amounts of changes in subsidy level.
|Characteristics||Sum of payment changes across five subsidy changes|
|$0 - $300 in sum of positive change in any payment change||$301 - more in sum of positive change in any payment change||X2; (significance) for block|
|3 or older||199||60|
|Education of adopting mother||High school or less||227||45||4.16^|
|Some college or more||265||80|
|Adopting parents have other (birth) children||No||301||73||.11|
|Employment of adopting mother||Outside of home||261||77||3.42^|
|$26,442 or less||131||29||9.39*|
|$48,762 or more||120||40|
|Age of adopting mother||20s or younger||80||13||4.98|
|50s or older||49||10|
|Number of minor children||1||94||21||2.38|
|2 - 3||235||72|
|4 or more||179||41|
|Single parent||Single parent||101||26||.015|
|^ p < 0.10
* p < 0.05
Family income and maternal education are associated with subsidy increases
We compared demographic differences in smaller amounts ($0 to $300) of overall increases and larger ($301 or more) amounts of monthly subsidy increases. Children aged 3 or older were more likely to receive larger subsidy increases over time, but this finding was not statistically significant. If adopted children lived with a well-educated adopting mother, they were more likely to experience high amounts of subsidy payment changes, X2;(1, 642) = 4.16, p < .05. Adopting mothers working outside the home prior to adoption were more likely to receive larger subsidy changes over time than mothers who were at home, although not at a statistically significant level (p = .065). In terms of family income, the families in the upper 50th percentile of family income were more likely to receive large amounts of subsidy changes over time than relatively low-income families, X2;(3, 642) = 9.39, p < .05. The association between the amount of payment change and children's race was very small and statistically insignificant (Exhibit 30).
The findings that families with mothers with higher education and families with greater affluence were more likely to receive higher amounts of subsidy increases are consistent with research suggesting that these parents have higher expectations for their children (Barth and Berry, 1991). In addition, these parents may be more able to advocate for subsidy increases.
If these findings hold up in the multivariate analysis that is, after controlling for the age of the child at the time of adoption (a proxy for child behavior problems) this would need to be considered in the evolution of a more equitable adoption subsidy program.
Multivariate analysis: logistic regression results. The bivariate results suggest that there are associations between children's and adoptive families' demographic characteristics and the amount of payment changes. We performed a logistic regression analysis in order to test associations between individual demographic characteristics after controlling for their association with other case characteristics and the amount of payment changes. We ran three slightly different models, each one including a somewhat different combination of variables, because all variables could not be tested simultaneously and because we wanted to see whether removing education or income which are highly correlated affected the results. Model 1 includes the child's race, age, and adopting mother's educational level; model 2 includes child's race, age, and adoptive family's income; and model 3 includes children's race, age, the adopting mother's educational level, and family income. The number of minor children, gender, single parent, and employment status were tried in the models, but did not improve the model fit and showed no significant relationship to the dependent variable.
A limitation of these models is the lack of data representing child characteristics such as disability and behavior problems, which should be strongly related to subsidy amount. Children's age serves to some extent as a proxy for these measures, since problems typically manifest as children grow older. The strongest predictor in these models is family income, although not entirely in the direction that would be expected if subsidies were being used to help families meet children's service needs.
As Exhibit 31 shows, all three logistic models appeared to be significant with acceptable, but not impressive, goodness-of-fit results. However, other results should be carefully considered because pseudo R² values are very small across all models (i.e., the models do not explain a sizable proportion of the differences in subsidy changes). In addition, of the independent variables, none of constituent item-level dummy variables -except family incomes less than $26,443 and family income of between $26,443 and $36,000 (at the time of placement in 1988- 89) were significant.
|Dependent Variable||Model 1||Model 2||Model 3|
|Sum of positive change in any payment change ($0 -$ 300, $301+)|
(Hosmer and Lemeshow Test)
|X2 = 7.77,
|X2 = 15.13,
|X2 = 8.38,
|Parameter Estimates||Odds ratios|
|Child's age||0 - 2||1.0||1.0||1.0|
|3 or older||1.33||1.35||1.35|
|Family income||< $26,442||N/A||1.00*||1.00|
|Mother's education||High school or less||1.00||N/A||1.00|
|Some college or trade school||1.27||N/A||1.23|
|Four year college or more||1.72*||N/A||1.57^|
|^p < .10
*p < .05
Event history analysis. We endeavored to understand the timing of payment changes in California in order to understand patterns of post-adoptive services need. Exhibit 32 shows the overall cumulative probability of the first payment change following placement. For this first payment change, and all subsequent payment changes, a large portion of AAP recipients has experienced payment change every two years because families must recertify their AAP status every two years. The portion of people with standard two-year payment changes who are receiving a fourth payment change or a fifth payment change is much smaller than the portion receiving routine payment changes at the first payment change or second payment change. That is, people who have experienced more payment changes are likely to more quickly experience other payment changes before two years.
Between the time of placement and the first payment change, only 25 percent of AAP recipients have experienced a payment change before the required two-year recertification. Yet about 41 percent of AAP recipients who have experienced a fifth payment change experienced their fifth payment change before two years from the date of fourth payment change. Exhibit 33 shows the quartiles for payment changes (estimated with Kaplan-Meier) and the proportion of the payment changes that occurred prior to the first year and prior to the routine second-year payment change.
|25th %||Median||75th %||< 1 year||< 2 years|
|Placement to 1st payment change||728 days||731 days||1035 days||11%||25%|
|1st payment change to 2nd payment change||609 days||731 days||N/A||17%||31%|
|2nd payment change to 3rd payment change||565 days||756 days||N/A||18%||31%|
|3rd payment change to 4th payment change||516 days||N/A||N/A||20%||34%|
|4th payment change to 5th payment change||245 days||730 days||N/A||32%||41%|
|Note: The 25th percentile indicates that 25 percent of families had a payment change at these times.
The 75th percentile indicates that 75 percent of families had a payment change by this time.
N/A indicates that the median and quartiles could not be estimated.
We next examined the probability of a payment change by the recipient's characteristics. Whereas there is little difference by child's age at adoption placement or educational level of adopting mother, the probability of payment change varies by family income and race. Confirming the logistic analysis, families with family incomes of $26,443 to $36,000 are significantly more likely to experience a payment change three years from placement. White, black, and Hispanic groups have similar "risk" of experiencing a payment change but children who are of "Other" races have a greater likelihood of experiencing a payment change.
We next endeavored to understand the likelihood of a payment change, simultaneously controlling for other case characteristics. Exhibit 34 shows median durations and risk ratios of a Cox proportional hazards model that analyzes the likelihood that a payment change will occur from placement and to fifth payment change while controlling for characteristics of adopted children and adoptive families. The median duration was two years (731 days) for each payment change.
These data clearly show that the timing of most payment changes was right at the two-year recertification point. Yet there were case characteristics that made the timing to payment changes vary significantly. Parents with at least some college, children who were three or older at the time of placement, and black children were the only groups whose risk ratios were significantly different from others, although this did not occur for all payment changes.
|Adoption placement to 1st payment change||1st payment change to 2nd payment change||2nd payment change and 3rd payment change||3rd payment change and 4th payment change||4th payment change and 5th payment change|
|Median duration||Risk ratio||Median duration||Risk ratio||Median duration||Risk ratio||Median duration||Risk ratio||Median duration||Risk ratio|
|3 or older||731||1.01||731||.98||731||1.30^||N/A||.90||638||1.34|
|$26,443 - $36,000||731||1.01||731||.87||731||1.11||N/A||1.02||N/A||.50|
|Mother's education||High school or less||731||1.00||731||1.00^||N/A||1.00||N/A||1.00||731||1.00|
|Some college/Trade school||731||.98||730||1.22^||731||1.24||N/A||1.06||730||1.00|
|Four-year college or more||731||1.14||730||1.28*||731||1.25||731||1.24||730||1.18|
|^ p < 0.10
* p < 0.05
Transitions to residential care. Because of the particular policy relevance of time-limited placements in residential treatment for children receiving subsidies insofar as the federal government will not reimburse for this, but 19 states will cover the costs (at least in part) we completed a model for those who had a payment change with a reason of "residential care." In an earlier report (Barth, Gibbs, and Siebenhaler, 2001), we had indicated that older children, white children, children in nonkinship adoptions, and children who were not in the deferred adoption agreement program were all more likely to receive residential care. In this analysis we examine some of these factors, and we also consider family income, mother's education, and the history of payment changes. We also learn about timing of those transitions to residential treatment.
Only 34 children in this sample entered residential care during the study time frame. This makes it impossible to estimate medians for individual variables. Yet a Cox proportional hazards model could be computed, and is shown in Exhibit 35. This model is consistent with earlier work showing that children adopted when older than three years have a higher likelihood of entering residential placement that is paid for by a payment change. (California does not pay for for-profit residential treatment, so some children may have entered residential treatment but not be included in these data.) The number of payment changes was also significantly related to a payment change for residential treatment. Although 11 children, about one-third of all entries to residential care, obtained a payment change for residential treatment as their first payment change, this was not typical. Most children who entered residential treatment had three or more prior payment changes. Parental income has a tendency to be related to a payment change for residential treatment, with the group of families earning between $36,001 and $48,761 at the time of adoption having the highest risk ratios. Neither race nor the education of the mother was significantly related to the use of subsidies for residential treatment.
|Risk ratio||95% CI Exp(B)|
|Education of adopting mother||High school or less||1.00|
|Some college or more||1.12||.52||2.41|
|Child's age at placement||0 -2||1.00|
|3 or older||2.07*||1.03||4.17|
|$36,001 - $48,761||2.67^||.86||8.26|
|Number of payment change||1 -3 of payment changes||1.00|
|3 or more payment changes||4.86*||2.25||10.50|
|^ p < 0.10;
* p < 0.05
The time to use a subsidy payment for residential care varied, in our study, from a little more than 2 years to 10 years, with the midpoint of those changes at about 7 years. This suggests that the likelihood of placement into AAP-funded residential care is accelerating.
"report.pdf" (pdf, 1.44Mb)