Understanding Foster Parenting: Using Administrative Data to Explore Retention

Developed by RTI International under contract to the
Office of the Assistant Secretary for Planning and Evaluation
U.S. Department of Health and Human Services
Contract: HHS-100-99-0006, delivery order #5

Project Director:
Deborah Gibbs
Research Triangle Institute

Project Officer:
Laura Radel
Office of the Assistant Secretary for Planning and Evaluation

"

Introduction

Foster homes are a critical resource within the child welfare system. In recent years, adoptions from foster care have increased dramatically, as has the use of relative caregivers for children in out-of-home care. Nevertheless, more than 260,000 children were in non-relative foster care at the end of FY 2001 (U.S. Department of Health and Human Services [DHHS], 2003).

In addition to maintaining sufficient licensed foster homes to house the children in care, child welfare agencies are challenged to provide foster care in placements that are stable, can accommodate sibling groups, and are located in proximity to family members (DHHS, 2000a). The increase in adoptions from foster care  from 37,000 in 1998 to 53,000 in 2002  creates an additional potential strain on foster home resources. Because the majority of adoptions are by foster parents, these homes may become less available as foster homes, following one or more adoptions. During the years for which national data are available, the percent of children who are adopted by their foster parents has ranged from 65 percent in 1998 to 59 percent in 2001 (DHHS, 2000; DHHS, 2001; DHHS, 2002; DHHS, 2003).

Foster parents thus play a central role within the child welfare system, both as resources in providing care that meets increasingly demanding criteria and as the primary resource for adoptive children. However, research on foster parent retention is surprisingly slender. Research related to foster parent retention typically describes the characteristics and experiences of foster parents based on their status (current or former foster parents) or their stated intention (to continue or cease foster parenting). Little is known, however, about the length of time actually served by foster parents and the characteristics that distinguish those with varying lengths of service.

The remainder of this section describes the objectives of this project and provides background information from previous research on foster parenting. Section 2 describes the administrative data and the methods for descriptive and multivariate analyses. Section 3 describes foster home characteristics and utilization, and Section 4 presents analyses of length of service for foster parents. Finally, Section 5 summarizes these findings and presents conclusions.

Project Objectives

This study was designed to extend current understanding of foster parent retention by producing unbiased estimates of length of service and examining factors associated with licensure, provision of care, and length of service. Principal research questions include

• How have the characteristics of foster parents changed over time?
• How can variations in activity levels be described, and what foster parent characteristics are associated with varying activity levels?
• What is the typical length of service for foster parents?
• What characteristics are associated with variations in length of foster parent careers?

An intermediate objective is to test the feasibility of using administrative data to describe foster parents, applying data management and analytic methods that have been used to describe the experience of children in foster care, including their length of stay (Wulczyn, 1996; Usher, Wildfire, and Gibbs, 1999).

Factors Associated with Foster Parent Retention

Three studies represent much of the recent research on foster parent retention. The National Survey of Current and Former Foster Parents, conducted in 1991, used a nationally representative sample to select more than 1,000 current and foster parents for interviews (DHHS, 1989). Data from this survey were the basis for more extensive descriptive analyses by Rhodes and colleagues (Rhodes, Orme and Buehler, 2001). In the second study, researchers at Ohio State University collected data from 539 current and 265 former foster parents in eight urban counties, using logistic regression to identify factors that distinguish ongoing from former foster parents (Rindfleisch, Bean and Denby, 1998) and predict intent to continue foster parenting (Denby, Rindfleisch and Bean, 1999). In addition, a recent assessment by the Office of the Inspector General conducted both interviews with child welfare managers and focus groups with foster parents on issues affecting foster parents (DHHS, Office of the Inspector General [OIG], 2002). These three studies, and other less comprehensive ones, yield fairly consistent findings on factors that influence foster parent retention.

Measures used in these studies include willingness to continue foster parenting, intention to continue or not, and satisfaction with foster parenting, which has been shown to be associated with intention to continue (Denby, Rindfleisch, and Bean, 1999). Determinants of continued foster parenting can be categorized in terms of foster parents'experiences (i.e., interactions with child welfare agencies and with foster children) and their demographic and socioeconomic characteristics.

Foster Parents' Experiences

Interactions with the child welfare agency were the most commonly cited factors affecting foster parent retention. In the National Survey of Current and Former Foster Parents, agency-related issues, including unsatisfactory interactions with workers and agency insensitivity, were cited as a reason for quitting by 37 percent of former foster parents and 62 percent of those intending to stop foster parenting. While former foster parents also cited the lack of services as an issue, an analysis of data from this survey found that reported service needs did not vary significantly among current foster parents, former foster parents, and those intending to quit (Rhodes, Orme, and Buehler, 2001). It appears that the level of concern about service needs, rather than the actual service needs, is related to foster parent status.

Former foster parents were three times more likely to be dissatisfied with a child's caseworker than current foster parents (DHHS, 1989). Foster parents who intended to leave were more likely than continuing foster parents to report that workers did not communicate expectations clearly and treated foster parents as if they were in need of help themselves (Rindfleisch, Bean, and Denby, 1998; Denby, Rindfleisch, and Bean, 1999). Problematic interactions with child welfare agencies also include those surrounding allegations of abuse or neglect, and interactions with agency red tape (Rindfleisch, Bean, and Denby, 1998).

Dealing with difficult behaviors among foster children was the most frequently cited challenge of foster parenting among those interviewed in a Nashville-area study (Martin, Altemeier, Hickson, Davis, and Glascoe, 1992). Child-related problems were cited as a reason for quitting by 24 percent of former foster parents (DHHS, 1989), and were also associated with both satisfaction and intent to continue foster parenting (Denby, Rindfleisch, and Bean, 1999). Other stresses surrounding the relationship with the foster child included the difficulty of seeing children return to birth parents, interactions with birth parents, and having no say in the child's future (Martin et al., 1992; DHHS, 1989; Rhodes, Orme, and Buehler, 2001; Hornby, 1985).

Experiences with both pre- and post-licensure training appear to affect foster parent satisfaction and retention. Using data from a longitudinal study designed to examine the impact of preservice training, Fees et al. (1998) found that foster mothers who described the training as useful were more likely to find satisfaction in the role demands of foster parenting. Boyd and Remy (1979) found a significant association between training and license retention. Examining different groups of foster parents within their study population, they found that the effect of training was strongest for foster parents who are less assertive and involved in community activism. Compared to current foster parents, former foster parents and those planning to quit were less likely to report having received adequate training, particularly related to dealing with teens and children with special needs (Rhodes, Orme, and Buehler, 2001). The data used do not allow investigation of whether exiting foster parents actually received training of lower quality or whether negative feelings about foster care experiences influenced foster parents' assessment of the training.

Personal crises or changes in the foster parents' circumstances may precipitate exit from foster care. Issues such as age, foster parents' health, and marital crises were cited by 29 percent of former foster parents in the National Survey of Current and Former Foster Parents (DHHS, 1989). Data from the same survey showed that 28 percent of former foster parents, and 18 percent of those planning to quit, reported doing so because they planned to adopt (Rhodes, Orme, and Buehler, 2001). Foster parents who cited wanting to adopt but having been unable to do so as a motivation for becoming foster parents were more than twice as likely to leave foster parenting than other foster parents, possibly because they had adopted foster children (Rindfleisch, Bean, and Denby, 1998).

Low levels of financial support for foster parenting were cited as a reason for quitting by 8 percent of former foster parents and 27 percent of those planning to quit (Rhodes, Orme, and Buehler, 2001). In the same study, former foster parents were more likely than current foster parents to report that they could not afford the cost of caring for the child most recently in their care (DHHS, 1989). In an Oregon demonstration project in which foster parents were randomly assigned to receive enhanced subsidies and services, enhanced subsidy only, or standard treatment only, participants receiving additional stipends and supports had a dropout rate that was two-thirds less than that of the control group over a 2-year period (Chamberlain, Moreland, and Reid, 1992).

Foster care board rates may affect the supply of foster homes if foster parents find that the cost of providing for children's needs exceeds the available support (Simon, 1975). An analysis of data from the 1980 Survey of Foster Parents in Eight States found that adjusted foster care board rates predicted whether licensed foster parents had any children in their homes (Campbell and Downs, 1987). However, the adjusted board rate did not predict the number of children cared for in those homes that provided care.

The fact that licensed foster parents may not be actually providing foster care, as suggested by the study above, suggests another dimension in assessing the supply of foster home care, that of home utilization. The National Survey of Current and Former Foster Parents found that 35 percent of licensed foster homes surveyed had no children in the home at the time of the survey  these homes were more likely to be nonurban and white (DHHS, 1989). An assessment of foster parent recruitment suggested that general campaigns bring in homes that are unwilling or unable to care for the children who are most likely to be in care (DHHS-OIG, 2002). While these homes may still be licensed, the foster parents have in effect discontinued foster parenting. At the other end of the spectrum, Martin et al. (1992) found that 23 percent of the foster parents interviewed cared for half of the children in care in the participating homes.

Foster Parent Characteristics

The characteristics associated with exiting foster parents are not as well described as foster parenting experiences. Findings are less consistent for both demographic and socioeconomic characteristics than for the foster parent experiences described in the preceding section.

Older foster parents appear more likely to continue providing foster care. Older foster mothers were significantly more likely to continue foster parenting rather than quit and were more likely to actually provide care (Rhodes, Orme, and Buehler, 2001; Campbell and Downs, 1987). Denby and colleagues (1999) found that age of foster fathers, but not foster mothers, was associated with increased intention to continue foster parenting.

Foster parent race was not associated with satisfaction with foster parenting (Denby, Rindfleisch, and Bean, 1999; Fees et al., 1998). However, Rindfleisch and colleagues (1998), using the same data as Denby, did find that white foster mothers had a significantly higher probability of having quit foster parenting.

Studies that examine socioeconomic characteristics generally find that higher levels of employment and income are associated with increased likelihood of quitting foster parenting. Although Rhodes and colleagues (2001) did not find significant income variation among current foster parents, former foster parents, and those intending to quit, continuing foster parents are more likely than the other two groups to earn less than $25,000 annually (DHHS, 1989). Foster parents for whom foster parenting is a source of income, and those who are unemployed, are more likely to continue (Rindfleisch, Bean, and Denby, 1998; Campbell and Downs, 1987).

These analyses of why foster parents leave are primarily based on self-reported data from foster parents. Response rates for former foster parents were substantially lower than for current foster parents, suggesting possible nonresponse bias. Among studies of why foster parents continue or leave, information on how long foster parents serve is notably absent. Only two of the studies report the time in foster parenting for participating foster parents: a mean of 8.6 years in Martin et al. (1992) and 5 years in Rindfleisch, Bean, and Denby (1998). None compare length of service among different groups of foster parents.

Methods

This section describes the data on foster parents and child placements used for all analyses in this report and the analytic methods used to describe foster parent characteristics, foster home utilization and foster parent length of service.

State Data

Child welfare agencies in three states  New Mexico, Oklahoma and Oregon  contributed data for these analyses. Selection of these states was based on data quality and states' willingness to provide both data and ongoing consultation to the study team. Table 2-1 summarizes data characteristics from each state. States provided three types of data for non-relative foster care: foster parent licensure data, data on individual foster parent characteristics, and placement records for children.

Foster parent licensure records included license types, license start dates, and license end dates. Because many homes had multiple license types, an analytic variable was created to identify those providing regular (non-relative) foster care only, foster-adopt homes (licensed foster homes that have indicated an interest in adoption and have completed some of the requirements for adoptive placements), restricted non-relative placements (homes approved for care of specific children), and therapeutic foster care (homes providing higher levels of care to children with special needs), as shown in Table 2-1. Homes licensed for relative care only were excluded from analysis, as were placements of relative children regardless of the foster parents' license type. Thus, for these analyses, the restricted license category represents foster parents licensed to care for specific children who are not relatives. In Oklahoma, this license is known as kinship non-relative; in Oregon, it is classified as a special license, along with a relative foster parent license.

Data on individual foster parents included race, date of birth, and number of foster parents in the home. Location was coded by New Mexico as urban or rural; for Oklahoma and Oregon, analysts coded homes as metropolitan or nonmetropolitan based on U.S. Census coding of counties (U.S. Census Bureau, 2000). High levels of missing data precluded analysis of data on race for New Mexico and data on ethnicity for all three states. This omission is unfortunate, since Hispanic children represent the majority of children in out-of-home care in New Mexico and substantial populations in the other states (CWLA, 2002).

To facilitate analysis at the foster home level, and to allow inclusion of both single- and two-parent foster homes in the analysis, the study team recoded individual foster parent characteristics into home-level variables. As an example, race was coded as one or both parents Native American, one or both parents black, all foster parents white, and other (other race or homes in which one foster parent was black and one Native American). Age was converted into a similar home-level variable representing age at first licensure. Additional data fields from Oregon included income and employment status at time of initial licensure. Note that these fields represent foster parents' characteristics at the time of initial licensure, and may change over the course of a foster parenting career.

Data on children placed in the home included date of birth, race and special needs identifiers (New Mexico and Oklahoma only). To allow examination of whether or not caring for children who might be seen as more demanding was a factor in foster parents' length of service, analysts coded children as infants (less than 1 year old at time of placement), adolescents (aged 13 or above at time of placement or before the placement end date) and special needs (physical, mental, or behavioral conditions identified). Oregon's data included a field identifying placements that ended because the child was adopted by foster parents.

Additional Analysis Variables

Analyses of foster home utilization and length of service were based on the span of time during which children were placed in the foster home, rather than licensing dates. These analyses were limited to homes in which the date of the initial license was known to occur after the dates for which child placement records were available to ensure that all placements in the home could be identified. This restriction created entry cohorts of foster parents whose entire foster parenting career could be examined.

Episodes of active foster parenting were defined as the number of days between the beginning of the first placement of a child in that home and the exit date of the last child placed in the home or the end of the study period. A gap of more than 90 days without a placement in the home signaled the start of a second episode of active foster parenting. Across the three states, between 74 and 87 percent of homes had only one episode of active foster parenting during the years studied. Among homes that were without placements for at least 90 days, only a minority were likely to subsequently resume foster parenting. All analyses of foster home utilization and foster parent length of service were based on the first episode of foster parenting.

Analysts created two measures to describe the intensity of foster care provision: occupancy rate and new placement rate. The occupancy rate was defined as the number of placement days for all children in the home divided by the number of days of foster parenting. It is equivalent to the average number of children in the home on a hypothetical day. Because episodes of active foster parenting may have included one or more periods of up to 90 days with no placements in the home, these occupancy rates are lower than those reported based on the average number of children in homes currently providing foster care.

To describe variations in the extent to which foster parents dealt with different children over time, the new placement rate was calculated as the number of new placements in a home, divided by the number of days in the first episode of foster parenting. The resulting figure was multiplied by 365 to create an annualized rate. A home with six new placements during 2 years of active foster parenting would have a new placement rate of 3; whereas a home with six new placements during 6 months of foster parenting would have a new placement rate of 12. Note that this formula can yield very high rates for foster homes that care for children for very short periods of time. A home that provided care to four children for 2 days, and had no other foster placements, would have an annualized rate of 4 ÷ 2 × 365, or 730. Very high rates for some groups are likely to represent short time in foster parenting, rather than homes that care for hundreds of children annually.

Analyses

The study team conducted three types of analyses. First, a descriptive analysis examined the characteristics of foster parents over the years for which data were available. Characteristics of interest included the demographic characteristics described previously and license types.

A second set of analyses described the utilization of licensed homes. These analyses compared foster parents with different characteristics in terms of whether any children were cared for during the time the home was licensed, the average number of children in the home, and an annualized rate representing the number of new placements in the home. Foster parent characteristics were also used to describe the likelihood of providing care for infants, adolescents, and children with special needs.

Finally, the team used longitudinal analysis to model the length of service in the first episode of active foster parenting. These analyses produced measures of time that are less biased than those based on cross-sectional data (Usher, Wildfire, and Gibbs, 1999) because they make use of right-censored data in which the event of interest (in this case, exit from foster parenting) has not yet occurred. Life table analyses estimated the cumulative probability of exiting foster care within specified time periods and examined factors associated with varying lengths of service. Kaplan-Meier analyses provided estimates of median length of service for foster parents.

In addition to bivariate analyses of the relationship between foster parent characteristics and experiences, the study team tested multivariate models using Cox proportional hazard regression (Allison, 1995). These models yield hazard rates, which can be conceptualized as the likelihood of an event  in this case, exit from foster parenting  on any given day. Higher hazard rates indicate an increased probability of exit, hence, a shorter length of service. Because New Mexico data were limited in both years of data available and number of cases, multivariate models used data from Oklahoma and Oregon only.

Foster Home Characteristics and Utilization

This section describes foster parent resources in terms of changes in number of foster parents over time, utilization based on the number of children cared for and the rate of new placements within a home, and characteristics of homes that are most likely to care for specific types of children.

Number and Characteristics of Licensed Homes

Both Oklahoma and Oregon experienced net growth in foster home resources over the years studied, as shown in Table 3-1.⁽¹⁾  Oklahoma experienced a 27 percent net growth over 6 years, and Oregon more than doubled the number of foster parent licenses over the 20 years reported. However, this growth occurred in the context of substantial turnover. The average number of licenses ending during year, as a proportion of active licenses, was 26 percent for Oklahoma and 21 percent for Oregon.

A graphic presentation of these data suggests the substantial efforts required to replace exiting foster homes each year to maintain and increase the number of available homes, and the high proportion of new, less experienced homes. Figure 3-1, for Oklahoma, and Figure 3-2, for Oregon, show the high proportion of licensed homes that were available for only part of the year because the license began or ended during the year.

Figure 3-1. Changes in Licensed Foster Parents,
Oklahoma

Some foster homes may become unavailable because foster parents have adopted the children in their care. Figures 3-3 and 3-4 compare the number of ending licenses, adoptive placements, and adoption finalizations for Oklahoma and Oregon, respectively.⁽²⁾ These data are also shown in Tables A-2 and A-6 in Appendix A. Both figures show increasing exits from foster parenting during years in which the number of adoptive placements or finalizations increased. Although not all adoptions are by foster parents, the parallel trends suggest a possible relationship between increased exits from foster parenting and adoptions by foster parents. However, none of the states' databases allowed comprehensive identification of children who were adopted by foster parents, which would have supported analyses of the length of service for these foster parents.

Figure 3-2. Changes in Licensed Foster Parents,
Oregon

Figure 3-3. Ending Licenses and Adoptions,
Oklahoma

Note: Data on adoptive placements and finalizations provided by Oklahoma Department of Human Services.

Figure 3-4. Ending Licenses and Adoptions, Oregon

Note: Data on adoptive placements and finalizations provided by Oregon Department of Human Services.
Data for adoptive placements only were available for 1997-1999; data for finalized adoptions only were available for 2000-2002.

Data on the characteristics of licensed foster parents over the years studied are included as Appendix A. These analyses have been reported in detail to each state and are only summarized in this report. New Mexico data (Table A-1) do not include enough years of data to identify trends.

Oklahoma and Oregon showed contrasting patterns in the use of restricted licenses. As noted in Section 2.1, these foster homes provide care only to specific children. As defined programmatically in Oklahoma, and specified analytically for Oregon data, this category excludes relative foster care. In Oklahoma, these homes increased from 2 percent of all licenses in 1996 to 13 percent in 2001, after having reached a peak of 16 percent of licenses in 1999 (Table A-3). In Oregon, restricted licenses decreased numerically and as a proportion of all licenses, from 25 percent in 1990 to 16 percent in 2002 (Table A-7).⁽³⁾ Because the foster parenting careers of these homes may vary from those licensed for regular foster care, their representation within the larger population is of interest.

The two states also had somewhat different trends with respect to foster parent age. In Oklahoma, the greatest growth in foster parent resources was among younger foster parents, whereas those in the middle age range increased only slightly and older foster parent homes declined (Table A-4). In Oregon, the greatest growth was among homes in which all foster parents were between 30 and 55 years of age (Table A-8). This distribution may have implications for adoptions from foster care because just over 50 percent of adoptive mothers are under age 30 (Dalberth, Gibbs, and Berkman, 2004).

In both states, changes in foster parent racewere small (Tables A-5 and A-9).

Oregon data include data on employment status of foster parents at the time of licensure. Between 1983 and 2002, the proportion of homes in which all foster parents worked full-time rose from 22 to 39 percent (Table A-10). This trend parallels changes in the age distribution of Oregon's foster parents, which showed declines in the proportion of younger foster parents (who are more likely to be home raising young children of their own) and older foster parents (who are more likely to be partially or fully retired).

Foster Home Utilization

The study team used several approaches to examine utilization patterns for foster homes. First, they examined the proportion of licensed homes that had no recorded placements and the characteristics of these inactive homes. They next looked at the average number of children and of new placements in the homes with different characteristics. Finally, the study team examined the distribution of foster care provided across the population of foster parents.

Active and Inactive Homes

New Mexico had a substantial number of inactive homes, which had no placements during their entire period of licensure. Among 866 homes with license dates in 1998 or later, 24 percent had no placements. Homes that were licensed for foster care only, had older foster parents, or were located in rural areas were most likely to be inactive. Oklahoma had very few inactive homes (4 percent). The structure of Oregon data files did not allow examination of inactive homes.

Occupancy Rates

The mean occupancy rates were similar across the three states, between 1.5 and 1.6, as seen in Table 3-2. This rate suggests that the average home has between one and two foster children on a hypothetical day, although such homes may have no children for part of the year and several children at other times. In all states, the mean occupancy rate was substantially higher than the median, shown in the lower portion of Table 3-2. This distribution suggests that a relatively small group of foster parents have much higher occupancy rates, for example, 10 percent of homes in each state had an average of four children in the home during their first episode of foster parenting.

The table shows several variations in occupancy among different types of foster homes, although feware large. In Oklahoma, the mean occupancy rate was higher for homes with restricted licenses than for other license types. In Oregon, the opposite pattern was seen: homes with restricted licenses had lower occupancy rates than did other types of homes.

Occupancy patterns varied across states for foster parent age and race. Occupancy rate was somewhat higher in Oklahoma for homes in which all foster parents were over age 55, but slightly lower for similar homes in Oregon. White foster parents in both Oklahoma and Oregon had lower occupancy rates than did Native American or black foster parents.

New Placements Per Year

The annualized rate of new placements, representing the number of different children placed in the home, varied considerably across states, as seen in Table 3-3. Because homes with a very short time in foster parenting will appear to have very high new placement rates, homes in which the length of service was less than 90 days were excluded from this analysis. This restriction excluded 21 percent of homes in New Mexico, 14 percent of homes in Oklahoma, and 19 percent of homes in Oregon. The variations seen among different types of homes are similar to patterns seen when all homes were analyzed.

New placement rates are similar for homes in Oklahoma and Oregon, but much higher in New Mexico. Since New Mexico's occupancy rates are similar to the other states, the difference in new placement rate may reflect shorter lengths of stay in foster care or higher rates of placement moves for these children.

Similar patterns for license type and foster home locationwere seen across states. Within each state, foster parents with regular foster care licenses had higher new placement rates than did those with other types of licenses. Rural or nonmetropolitan homes had consistently higher rates of turnover than urban or metropolitan ones.

Variations by foster parent age and number of foster parents in the home were inconsistent. In New Mexico, older foster parents had substantially lower new placement rates than others, whereas differences among age groups were small for other states. New placement rates were highest for Native American foster parents in Oklahoma and for white foster parents in Oregon. Similar to the pattern seen for age, single-parent foster homes in New Mexico had higher turnover rates than two-parent homes, whereas differences were small for the other states.

In Oregon, the rate of new placements was lower in homes where all foster parents work full time and in homes where income was greater than the median.

Although these analyses do not allow examination of the relative contribution of foster parent characteristics to variations in placement rates, the large and consistent disparity by license type suggests that this may be the most significant factor. Regular foster care homes care for many more different children over time than do foster-adopt homes, restricted foster care homes, or therapeutic foster homes. When compared with the with occupancy rates in Table 3-2, this analysis suggests that differences in placement turnover may be far greater than differences in the number of children placed in a home at one time.

Types of Children Cared For

To examine the relationship between characteristics of foster parents and the children cared for, the study team examined the percentage of homes that cared for at least one infant, at least one adolescent, and at least one child with special needs. These analyses were conducted in order to assess whether length of service was influenced by the types of children cared for in the foster home. Patterns varied across states for each type of placement. These variations may reflect variations in the foster care caseload across states, as well as the ways in which states use their foster home resources. The following discussion highlights only the very few consistent patterns seen for each type of placement.

Table 3-4 shows substantial variations among states in care for infants by homes with different types of licenses. For all states, two-parent homes were more likely to care for infants than single-parent homes. Some consistencies are seen for Oklahoma and Oregon. In these two states, homes with regular foster care licenses were nearly four times more likely to have cared for infants than homes with restricted non-relative licenses. Homes with younger foster parents, and those in rural or nonmetropolitan locations, were more likely care for infants. New Mexico had very different patterns for license type and age.

Table 3-5 shows that although the percentage of homes caring for at least one adolescent was similar across states, the characteristics of these homes varied across states. Across all states, two-parent homes were more likely to care for adolescents than single-parent homes. In both Oklahoma and Oregon, homes with restricted licenses and those with older foster parents were most likely to have cared for adolescents. In both New Mexico and Oregon, homes with foster-adopt licenses were the least likely to have cared for adolescents, suggesting the challenges of finding future adoptive homes for these children.

Table 3-6 shows contrasting patterns in care for children with special needs in New Mexico and Oklahoma, the two states for which these data are available. In New Mexico, the homes most likely to have cared for at least one child with special needs are those in which foster parents are between ages 30 and 55, and those in urban counties. In Oklahoma, children with special needs are more likely to have been cared for in homes with older foster parents, and those in which at least one foster parent is Native American. For both states, homes with regular foster care licenses and those with twoparents are more likely to have cared for these children.

Distribution of Foster Care

Although mean occupancy rates provide a useful measure for comparing groups of foster parents, the provision of foster care is in fact distributed quite unevenly across the population of foster parents. Table 3-7 shows that among foster homes that had at least one placement, many provided very little foster care. Across the three states, between 13 and 21 percent of homes provided less than 90 days of foster care during their time in foster parenting. Note that this figure represents the days of care provided to all children, rather than the length of service (i.e., 90 days of care might consist of three children placed in the home for 30 days each). In addition, this simple count of days of care provided does not adjust for the length of time in foster parenting, as do occupancy rate and new placement rate. Compared to all homes, those providing less than 90 days of care were less likely to have foster-adopt licenses (in New Mexico and Oregon) and more likely to have regular foster care licenses (in New Mexico) or restricted licenses (in Oregon). Age, race and location were not different for homes providing less than 90 days of care, but these homes were more likely than others to have only one foster parent.

At the opposite end of the spectrum, a small proportion of foster parents provided a large part of all foster care. The most active 20 percent of foster parents provided between 60 and 72 percent of all foster care days. Within this group, the most active 5 percent of homes provided more than one-quarter of all days of foster parenting.

The finding that a small group of foster parents provide the majority of care is striking. Interpretation of this pattern is difficult without additional data to suggest whether low utilization of some homes is due to geographic distribution of foster parents, or foster parents' preferences for specific types of children. These distributions are almost certainly influenced by the choices made by child welfare workers who match children with homes.For example, workers may choose to place children with experienced foster parents who they know and trust, rather than in less experienced homes. Little is known about how workers choose homes for specific placements.

Length of Service

This section describes length of service in foster parenting using longitudinal analysis methods. Descriptive analyses compared length of service for foster parents with different characteristics. Multivariate analyses examined the relative contribution of these characteristics to the likelihood of exit from foster parenting.

Foster Parents' Length of Service

Across the three states studied, the typical length of service in foster parenting was less than many children's stay in foster care. Median length of service was approximately 8 months in both New Mexico and Oregon, and approximately 14 months in Oklahoma. By comparison, the median length of stay for a child entering foster care was 5 months in Oregon; the 75th percentile was 18 months (D. Webster, personal communication, September 30, 2004). Although children's episodes of care may include planned placement changes, children whose stay in foster care is greater than the median length of foster parenting service are at risk of disruptions due to foster parent exits.

Table 4-1 shows that more than one-quarter of Oklahoma foster parents care for children for less than 6 months, with only one-third remaining in service more than 2 years. New Mexico and Oregon show even shorter lengths of service, with only one-fifth of homes remaining in service more than 2 years. As noted in Section 2, the dependent variable for these analyses is the length of the first episode of active foster parenting, rather than length of licensure.

Bivariate analyses were used to describe length of service in terms of foster parent characteristics and the characteristics of children in foster care. Table 4-2 shows contrasting patterns across the three states in the relationship between length of service and foster parent license type, age, and race. For all three states, foster parents in urban or metropolitan areas had longer lengths of service than those in rural or nonmetropolitanareas, and two-parent homes had longer lengths of service than single-parent homes. Length of service in Oregon was shorter in homes in which all foster parents worked full time and in homes in which all foster parents were home full time. This may reflect lack of time and the demands of rearing the foster parents' own children in the former case, and greater age in the latter case. Length of service was slightly higher for foster parents with greater than the median income than those at or below the median income.

Table 4-3 shows more consistent relationships across states between length of service and the characteristics of foster care provided. For New Mexico and Oregon, higher occupancy rates were associated with longer length of service.⁽⁴⁾  Length of service patterns related to occupancy levels in Oklahoma were inconsistent. Foster parents who provided care for some infants, adolescents, or children with special needs had longer lengths of service than those who cared for no such children or those who cared exclusively for these children. Because the new placement rate is highly sensitive to variations in length of service, as discussed in Section 2.2, this measure was not used for these analyses.

While it might be expected that caring for infants, adolescents, and children with special needs would be particularly demanding, these child characteristics do not appear to influence foster parents' length of service, except to the extent that homes caring exclusively for these children also have shorter lengths of stay. These findings suggest that the relationship between foster parents' length of service and the types of children cared for is not a simple one.

Multivariate Models of Length of Service

For Oklahoma and Oregon, Cox regression models were used to examine the relationship between length of service and foster parents'characteristics and activity.⁽⁵⁾  As noted earlier, the dependent variable for these models is a hazard ratio that represents the likelihood of exit from foster parenting, thus identifying variations in length of service.

For most variables, the largest stratum was used as the reference group against which the relative likelihood of exiting foster parenting was estimated for other groups. In the Oregon model, high levels of missing data for income and employment status limited the number of cases available for analysis. Because a model excluding these variables yielded similar findings, the model with all variables is shown here.

Findings shown in Tables 4-4 (for Oklahoma) and 4-5 (for Oregon) were generally consistent with the bivariate analyses described in the previous section. In reading these tables, the key statistic is the hazard ratio in the third column. Hazard ratios less than one indicate reduced likelihood of leaving foster parenting, or greater length of service. Hazard ratios greater than one indicate increased likelihood of exit, or shorter length of service. The fourth column indicates the statistical significance of the hazard ratio as compared to the reference category for each variable.

Table 4-6 summarizes the models for Oklahoma and Oregon. In both states, younger foster parents had significantly higher hazard ratios, indicating a higher likelihood of exit from foster parenting, or shorter length of service. Foster parents in metropolitan areas and those caring for infants, adolescents, or (in Oklahoma) children with special needs all had longer lengths of service. In Oregon, higher income was associated with longer length of service, but length of service did not vary by employment status.

Some of the length of service pattern variations seen in the bivariate analyses in Table 4-2 were eliminated when controlling for all variables. The increased length of service for foster parents with restricted licenses in Oklahoma was not apparent after controlling for factors such as age. A linear test of the impact of each set of variables on the overall model found that race was not significant in either Oklahoma or Oregon (p = 0.1057 and 0.9283). Foster home composition was not significant in the multivariate model for either state, although two-parent homes appeared to have substantially higher length of service in bivariate analyses.

Perhaps the most striking finding related to length of service is a pattern seen in bivariate analyses for all three states and persisting in multivariate analyses for Oklahoma and Oregon. Higher occupancy was consistently associated with increased length of service. In addition, care for children who might be considered more demanding  infants, adolescents, and children with special needs  was also associated with longer length of service.

Summary and Conclusions

Although the three states examined here are diverse in many ways, several consistent patterns in foster parent dynamics, utilization, and length of service are seen in these analyses. Licensing datain Oklahoma and Oregon show consistently high rates of foster parent turnover in both states;at least one in five foster homes exited the system each year. Ongoing attrition of foster parents creates enormous demands on systems that must recruit and train sufficient numbers of new foster parents to maintain and even expand the number of available homes.

Patterns of foster care provision varied across sites, but some clear trends were evident. Regardless of their characteristics, foster homes had, on average, between one and two children in the home at a time. In general, homes with nonwhite foster parents, those in rural or nonmetropolitan counties, and those with two parents cared for more children at a time and had higher rates of placement turnover. Foster parents caring for infants were typically younger, urban, and in two-parent homes, whereas those caring for adolescents were likely to be older, rural, and in single-parent homes. Across the three states, one-fifth of the foster parent population provided between 60 and 72 percent of all days of foster care.

Median length of service in foster parenting ranged from 8 to 14 months across the three states, suggesting that many children's placements in foster care are longer than the typical foster parent career. Multivariate models showed that foster parents with greater length of service are likely to be older, live in a metropolitan area, and be engaged in more intense foster parenting activity, as indicated by higher occupancy rates and care for infants, adolescents, and children with special needs. In contrast to earlier research, higher income was associated with longer length of service among Oregon foster parents; it is not possible to tell whether this is a distinct pattern for that state or a result of different methodologies. Whereas earlier research found longer tenure among black foster parents, this study found no significant associations between length of service and race after controlling for other variables.

Key findings from this study address multiple aspects of the dynamics of foster parent utilization and retention:

• Length of service in foster parenting is shorter than many managers would expect. As with analyses of children's length of stay in foster care, estimates of foster parents' length of service based on longitudinal methods may be surprisingly short. Cross-sectional samples and anecdotal data are both likely to overrepresent the long-term foster parent, who is disproportionately likely to be present when a survey is mailed, a focus group convened, or statistics compiled from rosters. Estimated with a longitudinal model, the median length of service of 8 to 14 monthsfor these states is a distinct contrast to the mean time in foster parenting of 5 to 8 years reported in the studies described in Section 1 (Martin et al., 1992; Rindfleisch, Bean, and Denby, 1998). In the three states studied, between 47 and 62 percent of foster parents exit foster parenting within a year of the first placement in their home.
• Foster parent burn-out cannot be identified as a factor in length of service. A working hypothesis at the outset of this study was that foster parents exit the system after being exhausted by high levels of placements in their homes and the demands of children in their care. This theory was not supported by the data. Higher foster home occupancy and higher levels of care for infants, adolescents, and children with special needs were consistently associated with greater lengths of service.
• One-fifth of the foster parent population provides 60 to 80 percent of all foster care. Across the three states, a relatively small group carries much of the work of the foster care system. These foster parents are similar to those described by Martin et al. (1992), who found that 23 percent of foster parents interviewed cared for half the children in care at the time. These foster parents may represent a core group of active and experienced foster parents. Because they are willing to accept a variety of placements, and because their long tenure equips them with practical expertise in caring for the children in need of placements, child welfare workers are likely to feel confident placing children in these homes, therefore, these homes carry a major portion of the workload.

Readers should note some important limitations of these analyses. First, the experiences of three states cannot be generalized to other groups of foster parents. Our analyses identified some consistencies among states, such as the uneven distribution of the foster parenting workload and increased length of service among foster parents who are over age 30, located in a metropolitan area, and caring for more children at a time. However, findings varied sharply among states for many key measures, such as the median length of service. It is not possible, based on analyses with three states, to speculate about which patterns may be more typical of foster parents in general.

A second limitation is that these analyses, while describing length of service and associated foster parent characteristics, provide little insight regardingwhy foster parents stay or leave. Rather, they offer a useful counterpoint to the research described in Section 1.2 on how foster parents' perceptions and experiences influence their decision to continue foster parenting.

A final limitation is that these analyses focus on associations between foster parent characteristics and the number and types of children cared for. However, child placements ultimately rest on child welfare workers' decisions, as well as foster parent preferences. These dynamics are likely to be far more subtle than can be revealed by examination of administrative data.

These analyses extend previous research by providing unbiased estimates of length of service for foster parents, as well as a more detailed picture of the characteristics associated with varying length of service. Further analyses in other states might build on these analyses to incorporate data elements such as foster parent training and foster care board rates.

For individual foster parents, the decision to continue or leave foster parenting is no doubt influenced by experiences with child welfare agencies and foster children and personal circumstances, as described in Section 1. Though longevity is of course not the only goal for foster parents, preventing the unnecessary loss of qualifiedfoster parents would significantly enhance child welfare systems' ability to enhance the safety, permanency, and well-being for children in their care. Better understanding of foster parent length of service and service dynamics is an essential first step toward achievement of this goal.

References

Allison, P.D. (1995). Survival Analysis Using the SAS System. Cary, NC: SAS Institute.

Boyd, L.H., and Remy, L.L. (1979). Foster Parents Who Stay Licensed and the Role of Training. Journal of Social Service Research, 2(4), 373-387.

Campbell, C., and Downs, S.W. (1987). The Impact of Economic Incentives on Foster Parents. Social Service Review, December, 599-609.

Chamberlain, P., Moreland, S., and Reid, K. (1992). Enhanced Services and Stipends for Foster Parents: Effects on Retention Rates and Outcomes for Children. Child Welfare, 71(5), 387-401.

Child Welfare League of America. (2002). Special tabulation of the Adoption and Foster Care Analysis Reporting System. Retrieved October 7, 2004 from http://ndas.cwla.org.

Dalberth, B., Gibbs, D., and BerkmanN. (2004). Understanding Adoption Subsidies: An Analysis of AFCARS Data, Draft Report. Research Triangle Park: RTI International.

Denby, R., Rindfleisch, N., and Bean, G. (1999). Predictors of Foster Parents' Satisfaction and Intent to Continue to Foster. Child Abuse & Neglect, 23(3), 287-303.

Fees, B.S., Stockdale, D.F., Crase, S.J., Riggins-Caspers, K., Yates, A.M., Lekies, K.S., et al. (1998). Satisfaction with Foster Parenting: Assessment One Year after Training. Children and Youth Services Review, 20(4), 347-363.

Hornby. (1985). Why Foster Parents Quit. Portland, ME: University of Southern Maine.

Martin, E.D., Altemeier, W.A., Hickson, G.B., Davis, A., and Glascoe, F.P. (1992). Improving Resources for Foster Care. Clinical Pediatrics, July, 400-404.

Rhodes, K.W., Orme, J.G., and Buehler, C. (2001). A Comparison of Family Foster Parents Who Quit, Consider Quitting, and Plan to Continue Fostering. Social Service Review, March, 84-114.

Rindfleisch, N., Bean, G., and Denby, R. (1998). Why Foster Parents Continue and Cease to Foster. Journal of Sociology and Social Welfare, 25(1), 5-24.

Simon, J.L. (1975). The Effect of Foster-Care Payment Levels on the Number of Foster Children Given Homes. Social Service Review, September, 405-411.

U.S. Census Bureau. (2000). Population Estimates for Metropolitan Areas and Components, Annual Time Series April 1, 1990 to July 1, 1999 (includes April 1, 1990 Population Estimates Base). Retrieved March 9, 2004, from http://www.census.gov/population/estimates/metro-city/ma99-03a.txt

U.S. Department of Health and Human Services (DHHS). (1989). The National Survey of Current & Former Foster Parents. Washington, DC: National Clearinghouse on Child Abuse and Neglect Information.

U.S. Department of Health and Human Services (DHHS) (2000). The AFCARS Report: Interim Estimates for Fiscal Year 1998  April 2000(3). Retrieved March 31, 2004, from http://www.acf.hhs.gov/programs/cb/publications/afcars/ar0400.pdf

U.S. Department of Health and Human Services (DHHS). (2000a). Child and Family Services Reviews Procedures Manual. Retrieved March 31, 2004, from http://www.acf.hhs.gov/programs/cb/cwrp/tools/index.htm

U.S. Department of Health and Human Services (DHHS) (2001). The AFCARS Report:  Interim FY1999 Estimates as of June 2001(6). Retrieved March 31, 2004, from http://www.acf.hhs.gov/programs/cb/publications/afcars/june2001.pdf

U.S. Department of Health and Human Services (DHHS). (2002). The AFCARS Report:  Interim FY2000 Estimates as of August 2002(7). Retrieved March 31, 2004, from http://www.acf.hhs.gov/programs/cb/publications/afcars/report7.pdf

U.S. Department of Health and Human Services (DHHS). (2003). The AFCARS Report:  Preliminary FY2001 Estimates as of March 2003(8). Retrieved March 31, 2004, from http://www.acf.hhs.gov/programs/cb/publications/afcars/report8.pdf

U.S. Department of Health and Human Services (DHHS)  Office of the Inspector General. (2002). Retaining Foster Parents (Publication No. OEI-07-00-00601). Retrieved May 3, 2004, from http://oig.hhs.gov/oei/reports/oei-07-00-00601.pdf

Usher, C.L., Wildfire, J.B., and Gibbs, D.A. (1999). Measuring Performance in Child Welfare: Secondary Effects of Success. Child Welfare, 78, 31-51.

Wulczyn, F. (1996). A Statistical and Methodological Framework for Analyzing the Foster Care Experiences of Children. Social Service Review, June, 318-329.

Endnotes

1. Trends in licensed homes were not examined for New Mexico due to limited years of data and incomplete data for 1998.

2. Adoptive placements are placements of children with parents who intend to adopt them. Adoption finalizations are the legal completion of adoption arrangements.

3. The categories used for licensing foster parents changed in 1990, making it difficult to compare data from 1983  1989 with more recent data.

4. Different occupancy strata were used for analysis in Oregon because its median occupancy rate was 1.0.

5. Multivariate analysis was not done with New Mexico data due to data limitations for that state.

Acknowledgements

This report is based on a task titled Dynamics of Foster Parenting, which was conducted by RTI International, under contract number HHS-100-99-0006, delivery order 5, for the Office of the Assistant Secretary for Planning and Evaluation, U.S. Department of Health and Human Services. The opinions expressed in this paper are those of the authors and do not necessarily represent positions of the U.S. Department of Health and Human Services.

Essential to the production of this report was the cooperation of the state child welfare agencies in New Mexico, Oklahoma, and Oregon. Staff from these agencies provided extensive expert advice on the correct use of their state's data.

Material contained in this publication is in the public domain and may be reproduced, fully or partially, without permission of the Federal Government. The courtesy of attribution is requested. The recommended citation follows:

Gibbs, D. Understanding Foster Parenting: Using Administrative Data to Explore Retention (Washington, DC: U.S. Department of Health and Human Services, Office of the Assistant Secretary for Planning and Evaluation, 2005).

How to Obtain a Printed Copy

To obtain a printed copy of this report, send the title and your mailing information to:

Human Services Policy, Room 404E
Assistant Secretary for Planning and Evaluation
U.S. Department of Health and Human Services
200 Independence Av, SW
Washington, DC 20201

Fax:  (202) 690-6562