Printer Friendly

Attitudes toward out-of-home care over 18 months: changing perceptions of youths in foster care.

This article seeks to uncover children's evolving views of placement and to delineate characteristics associated with positive and negative attitude change over time. The authors used a subsample drawn from the National Survey of Child and Adolescent Well-Being. The subsample of 290 youths age seven and older who had been in out-of-home placement during the first 18 months of this study was analyzed by using latent transition analysis (LTA) to understand how children's views of placement changed over time and what covariates might be associated with these changes. A three-class solution was the most stable at both time points and highlighted three different sets of perceptions about out-of-home care. The LTA provides additional detail demonstrating that the majority of youths did not change their attitudes over the 18 month time span, particularly those who are very happy or very unhappy in their current placements. Age, gender, and mental health status were related to particular types of transitions. The results demonstrate the heterogeneity of experiences for youths in out-of-home care and highlight the need for tailored policies and interventions to assist youths in processing such experiences.

KEY WORDS: foster parents; latent transition analysis; out-of-home care; youth attitudes

**********

Clearly, out-of-home care benefits many children who are not safe living with their biological caregivers. Research from the 1970s forward documents developmental gains for children in care compared with those who return home (Fanshel & Shinn, 1978; Horwitz, Balestracci, & Simms, 2001; Leitenberg, Burchard, Healy, & Fuller, 1981; Wald, Carlsmith, & Leiderman, 1988). Recent findings confirm that children who remain in foster care for six years exhibited fewer problem behaviors than those reunified with their parents, despite having multiple placement moves (Taussig, Clyman, & Landsverk, 2001) and that contact with social services allows children in need of mental health services to obtain them (Farmer et al., 2001). Yet, an underlying anxiety common to policymakers and frontline child welfare workers alike is that the documented developmental gains achieved in foster care come with a high emotional price: the pain of separation from one's family of origin.

Prior research has used snapshots of children's experiences in care or retrospective descriptions by foster care alumni to understand the effect of out-of-home placement. This literature supports the contention that children are generally satisfied with their out-of-home placements. Studies using both small and large samples found high satisfaction with caregivers and few reports of serious problems (Johnson, Yoken, &Voss, 1995; Wilson & Conroy, 1999). Indeed, a sample of Canadian foster children consistently rated their foster families as emotionally "healthier" than their biological families (Kufeldt, Armstrong, & Dorosh, 1995).

Yet, the experience is not uniformly positive. Early findings from the most recent Casey National Alumni Study show a birthrate to teenagers in care to be double the national rate (17.2% compared with 8.2%). Homelessness affects more than one-fifth of youths for at least one night in their first year following discharge from care (Casey Family Programs, 2003). Many foster children retrospectively report concerns about their educational experience while in care (Barth, 1990; Festinger, 1983; Wedeven, Pecora, Hurwitz, Howell, & Newell, 1997), the severity of punishments in their foster homes (Fanshel, Finch, & Grundy, 1990), and a desire to have more influence in decisions about placement and visitation with their parents (Festinger, 1983). Furthermroe, the clinical literature associates severe behavioral issues such as suicide attempts with competing loyalties between foster and biological families (Haight, Black, Workman, & Tara, 2001; Pilowsky & Kates, 1996). Idealization of biological parents precipitated by placement is also viewed as problematic (Kufeldt et al., 1995), with some workers describing a need to counter children's" unrealistic fantasy of the perfect family" (Peters, 2005, p. 599). The separation experience, both at the time of removal and in the months and years following, leaves most children feeling sad, depressed, or upset (Fanshel & Shinn, 1978; Johnson et al., 1995). Most children reported that they missed their parents most of the time in the months and years following removal (Johnson et al., 1995).

Furthermore, the literature examining placement changes reports that small (8%) but significant numbers of youths change placement because of foster family issues, suggesting a need for more support for some foster parents. Children with mental health problems appear to have more difficulty in care, indicating that coupling mental health services with out-of-home placement may be needed to create a strong alliance between a foster family and a youth (James, 2004).

A recent cross-sectional investigation that used representative national data provides reassurance that most children in care are happy with their current caregivers (Chapman, Wall, & Barth, 2004).Yet, even in this generally positive picture of foster care, a majority of youths expressed ambivalence about care, saying that they felt connected to their current caregivers but still hoped to return home or at least have more connection to their biological families.

This mixed picture raises questions about how children make sense of foster care even as they benefit from increased stability and safety. In addition, more and more researchers are seeking to answer the question "What works for whom?" pointing out that within the general picture of success or failure of a particular program or intervention, there is often considerable heterogeneity of experience. For example, Gibson (2003) illustrated the importance of examining subgroups within an anti-poverty effort to determine what cluster of services seemed to have the most benefit for people with particular demographics. In her analysis, Gibson pointed out the danger of simply comparing participants with nonparticipants even when random assignment is present. The nuances of program success or failure are likely to be overlooked, and valuable lessons about which intervention components are most effective for which participants may be lost.

Understanding children's adjustment to out-of-home care presents a different, yet related, challenge. Children are not randomly placed into foster care and each foster home is different, meaning that the "intervention" of foster care may look very different for each individual child. Some children will be placed in homes with older foster parents, some will have foster siblings in the home or be placed with their own siblings, some will be a single child in a home, and some will be one of many. The possibilities of what a specific situation may look like for a particular child are almost endless. Adding to the complexity are the specific characteristics and experiences of each child that enters care. Yet, in each placement situation, some emotional adjustment must be made for a child and a foster family to connect and thrive.

Currently, the literature tells two stories, both of which are likely true: (1) out-of-home placement is difficult for children and (2) out-of-home placement benefits children. Direct service providers and policymakers are left with difficult choices to weigh. This balancing act may be eased by knowledge about the process of adjustment to out-of-home care: how a young person's attitudes change over time and what factors are associated with these changes. This article seeks to "uncover" these processes. With clustering statistical techniques, specifically latent transition analysis (LTA), children's evolving views of placement can be seen. Covariates are combined with measures of contentment in care to create the classes of children, and transitions between classes over time are estimated. Following the transition analysis, additional variables of interest are regressed on the probabilities of specific transitions to understand the relationship between these variables and the particular ways in which feelings evolve over time (the latent transitions).This type of analysis has not heretofore been possible because of limitations in child welfare data and because of the limitations of the analytic software available. However, with the use of data from The National Survey of Child and Adolescent Well-being (NSCAW), a longitudinal, probability study authorized by the Personal Responsibility and Work Opportunity Reconciliation Act of 1996 (PL 104-193), together with person-centered analytic techniques, distinct classes of children evincing positive and negative attitudes toward care can be seen, and any change in attitude can be documented over time.

METHOD

Sample

NSCAW documents the characteristics, needs, experiences, and outcomes for children and families involved in the child welfare system. The target population of NSCAW is "all children in the U.S. who are subjects of child abuse or neglect investigations (or assessments) conducted by CPS and who live in states not requiring agency first contact" (Dowd et al., 2004, p. 16). Thus, all children in NSCAW enter care because of reports of child maltreatment. Children entering the child welfare system through other means were ineligible for the sample. NSCAW has a multistage, stratified, cluster sampling design. Children were selected from 92 primary sampling units in 97 counties nationwide. The NSCAW design oversamples infants, children receiving services, children living out-of-home, and children who are victims of sexual abuse. Sample weights allow for the sample to be generalized back to the target population.

Although the entire NSCAW cohort includes 5,501 children, this study draws on data from 290 children who were in out-of-home care at baseline and at 18 months post-baseline and who were over six years of age at the first data collection point. These children represent 66,023 children in the study population. The analysis sample contains 134 males and 156 females representing 43.5% and 56.5%, respectively of the out-of-home child welfare population over six years of age. The sample is made up of 124 African Americans, 109 white Americans, 37 Hispanics, and 20 others. White Americans and African Americans represent nearly the same proportion of the study population, about 41% and 42%, respectively. Hispanics represent 12.1% of the study population. The majority of the sample, 172 children, is over 11 years of age, representing 57% of the population. Approximately 54% of the population represented by this sample has a score on the Child Behavior Checklist (CBCL) (Achenbach, 1991) at baseline that is in the borderline or clinical range.

Measures

Two sets of measures are used: One set of measures is used to create the initial latent classes and define the transitions present in the data; the second set are measures of specific child and placement characteristics that we model with regression analysis to understand what characteristics are associated with particular transitions.

Latent Class Creation Measures

A series of single-item questions from the University of California at Berkeley Foster Care Study (Berrick, Frasch, & Fox, 2000) were used that asked children to describe the experience of being in care, including perceptions of their current placement and their beliefs about their families of origin. These items were extensively pretested and modified before use (Berrick et al., 2000). Six questions (see Table 1 for the specific items and their population statistics), asked at two points in time, were used to develop the latent classes at both time points. These questions delineate three domains: (1) the child's perceived ability to stay in the current placement, (2) the child's beliefs regarding reunification with the biological family, and (3) the child's desire for permanency in the current placement. Furthermore, a dichotomous measure of gender was used as a covariate in creating the classes, as was a dichotomous measure of mental health problems, using the total problem score at the first time data collection of the CBCL (Achenbach, 1991), a widely used instrument for assessing mental health functioning. This score was dichotomized into children who had scores in the normal range and children who had borderline or clinical scores at baseline.

Measures Used in the Regression Analysis

Five measures were used in the regression analyses. The first was a dichotomous measure of age, categorizing children as seven to 10 years old or 11 years and older at the time of the initial investigation. A dichotomous race measure of white compared with nonwhite was created. The measure of gender was also included. A measure of mental health symptoms at the 18-month interview was created by using the CBCL total problem score. Children were classified as being in the normal rather than the borderline or clinical range. Finally, a measure of placement stability was created using the total number of placements experienced between the two data collection points. Children were considered stable if they had experienced only one placement during the time frame and unstable if they had experienced two or more placements.

Analysis

Within a group with similar experiences, it is sometimes difficult to capture the heterogeneity of attitudes and reactions to that experience. Person-centered analytic techniques address this issue by examining clusters of individuals within a larger group who vary in systematic ways both cross-sectionally and over time. Person-centered techniques seek an underlying structure in a given set of data. Accordingly, specific hypotheses are not tested, although prior research and theory are the basis for choosing the questions around which the classification will be built.

In the current analysis, two person-centered techniques were used. Latent class analysis (LCA) was used to define groups of children into classes on the basis of their response patterns to the six questions in Table 1. In LCA a latent nominal variable is derived from multiple observed dichotomous or nominal indicator variables (Heinen, 1996; Lazarsfeld, 1950). The latent variable is an unobserved variable that categorizes children into classes (or groups) representing various profiles. In the current analysis, the latent variable comprised various indicators presented in Table 1, which form the latent construct contentment in out-of-home care. With this method, we obtain the probabilities of a random member of the study population belonging to each of the estimated classes as well as the conditional probability of giving a certain response to each of the indicator variables given membership in a specific latent class. The result is a set of classifications in which youths whose answers to these questions are similar are grouped together.

Once these classes were established, we used a latent transition model to assess changes in class membership between the baseline survey and the 18month (post-baseline) survey. LTA allows for the observation of change over time in latent class membership (Collins, Graham, Rousculp, & Hansen, 1997; Collins & Wugalter, 1992; Graham, Collins, Wugalter, Chung, & Hansen, 1991). LTA estimates transition probabilities of moving from one latent class or status to another at a later time point. With LTA, we obtained estimated probabilities of a random member of the population making each possible transition given the same latent classes at each time point. Measurement of latent classes was held constant across time points by specifying the same number of classes and constraining threshold estimates at each time point to be equal. When an acceptable transition model had been established, youths moved from probable membership in a particular latent class to occupying a "latent status," the difference being that the latent status describes the way in which one's attitudes toward care may or may not change over time. Thus, a child who is a member of a "want to go home" class at time 1 but who reports contentment in out-of-home care and a desire for adoption at time 2 may occupy the latent status of a "positive changer" one whose attitude toward care becomes more positive over time. Covariates were incorporated at each point to enhance the ability to create stable classes over time points. The influence of the covariates was used to aid in classification and in the transition analysis.

All analyses were estimated using the Mplus Version 4.1 statistical package (Muthen & Muthen, 1998-2006). Parameter estimates and standard errors were estimated by using probability weighting and accounting for the stratification and clustering resulting from the complex sampling design of NSCAW. Maximum likelihood estimation for missing data was used (Arbuckle, 1996), which allowed the retention of all cases that might have been missing on one or more latent class indicator variables.

A final analysis step involved conducting a series of regression analyses to model the relationship between probability of membership in a particular transition status and specific child characteristics. Thus, the child probabilities for membership in each transition status were regressed on child characteristics. The variables of interest were then related to transition group membership.

RESULTS

Classes are fit to the data at the two time points separately with gender and behavior problems included as controls or "grouping variables," which help classify children in the LCA. In this analysis, within-time latent class fitting was used primarily to explore and describe the latent class variable, contentment in out-of home care, at both time points. LCA was also used to choose the number of classes represented by the data. This was done by fitting LCA models with different numbers of classes and comparing the models by evaluating model fit indices while also considering the substantive meaning of the classes in each model. Some authors liken this to factor analysis in which the number of factors is decided upon by using both statistical cut-off points and the author's substantive knowledge to identify meaningful differences between the item groupings. A summary of the fit statistics for various latent class analyses done at the first and second time points is provided in Table 2.

As Table 2 demonstrates, a three-class solution is the most stable at two time points. Determining the final number of classes is an iterative process in which models with different numbers of classes are compared. Together with a series of fit indices, the investigator's ability to interpret various solutions informs the choice of a final model. Two widely recognized indicators of a strong model are the Bayesian Information Criteria (BIC) statistic and the entropy measure. To make a determination an investigator runs a series of models beginning with a one-class solution. The BIC values are compared as the number of classes in the solution increases. A decreasing BIC value is one indication of a better fitting model. Conversely, the entropy measure, which describes how well a model discriminates groups, should be close to one. In this current analysis, the decreasing BIC statistic and the increasing entropy measures were used, together with interpretability of the model, to choose a three-class model. Although the BIC decreases at both time points for a four-class solution, the entropy measure begins to decline at the second time point in the four-class solution. Furthermore, when the models were examined for substantive differences, none could be found.

In the best case scenario, classes at time 1 will mirror those at time 2. This stability indicates that legitimate classes have been identified and that participants may move into and out of these classes. We subsequently used the three class latent variables in a transition model. This model is depicted in Figure 1. Latent classes are conditioned on gender and baseline CBCL at both time points, and the transition probabilities are estimates through the path from the latent class variable at time 1 to the latent class variable at time 2 (18 months later).

A description of the three classes based on responses to the six latent class indicators in the transition model is presented in Table 3. These are probabilities of a "yes" response conditioned on membership in a given class. Members of the first class want to return home. Members of the second class believe they can stay in their current placement, and many may want to stay but also hold out hope that "things would be different" if they were to return to their biological families. Those in the third class appear committed to their current placements and clear that returning home is not something they want or expect to happen. Accordingly, we refer to the classes as "want to go home," "happy but hopeful" and "content in care." At time 1, most of the children, approximately 58% of the study population, had the highest probability of being members of the want to go home class. Members of the happy but hopeful and content in care classes, on the basis of the highest probability of membership, make up 26% and 16%, respectively, of the study population. The distribution at time 2 changed to include 42% in the want to go home class, 31% in the happy but hopeful class, and 27% in the content in care class. The three-class transition model has good entropy (0.901), which means that children had high probabilities of being in one of the three classes. This indicates that the model does a good job of classifying children into classes.

[FIGURE 1 OMITTED]

The transition probabilities presented in Table 4 give the conditional probability of moving from each class at time 1 to a different class at time 2 (18 months later). Each cell of the table represents one of nine potential latent statuses. Two possible transitions or statuses did not occur in the data: transitioning from the content in care class to either the happy but hopeful class or the want to go home class. The transition probabilities show that 62.9% of children in the want to go home class, 55.4% of children in the happy but hopeful class, and 100% of children in the content in care class remained in the same class over time. Of the children who changed classes, the highest probability was for those in the want to go home class to move into the happy but hopeful class (28.5%).And 24.9% of those in the happy but hopeful class at time 1 moved into the content in care class at time 2. Others moved from want to go home to "content in care" (8.7%) and from happy but hopeful to want to go home (19.6%) over time. In general, the probabilities were highest for moving from less to more contentment in care.

We may also consider the overall proportion of the study population in each combination of classes. The majority of youths, 36%, were in the want to go home class at both time points. Approximately 15% remained in the happy but hopeful class at both time points, and 16% remained in the content in care class at both time points. Of the children who changed class membership over time, 17% moved from wanting to go home to happy but hopeful; 5% moved from wanting to go home to content in care; 5% moved from happy but hopeful to wanting to go home; and 6% moved from happy but hopeful to content in care. Therefore, although the highest probability in terms of movement was toward contentment in care, most of the children in the study population did not change classes in this time period, and most maintain some level of desire to return to their biological families.

The percentage of the total unweighted sample likely to occupy each status is presented in Table 5. These figures reiterate that for those youths whose attitudes changed, most moved to more positive views of out-of-home care.

Covariate Analysis

The CBCL score at time 1 did not have a significant impact on creating the classes within time. Being a girl decreased the probability of being in the want to go home and happy but hopeful classes compared with the content in care class at time 2. No other significant associations were estimated in the transition analysis in terms of latent class grouping variables.

Regression Analyses

A variety of associations were noted between child characteristics and specific transition statuses. The only characteristic that did not appear to be related to transition status was race (see Table 6).Age, placement stability, CBCL scores at both time points, and gender were related to particular types of changes in attitudes toward care. Younger youths had an 18% higher probability of remaining content in care over time compared with older children. Children with stable placements had a 13% higher probability of remaining happy but hopeful about returning home over the two time points compared with children with unstable placements. Being in the nonclinical range at the second measurement point made it 23% less likely that one would remain in the want to go home group and 14% more likely to move from the want to go home category to a happy but hopeful outlook. Finally, gender was associated with all transitions, with boys more likely than girls to move or stay in more positive statuses. Coefficients for these findings are presented in Table 6.

DISCUSSION

Before a discussion of the LCA and LTA results, it is useful to consider the information contained in Table 1, which describes the percentages of youths answering "yes" to each of the questions used to create the classes. These numbers give a sense of the overall group's view of out-of-home care at two time points. Through this lens, it appears that youths generally become more positive about care over time and less positive about returning home, as indicated by the rising percentages of positive responses to the first two questions and decreasing positive responses to percentages on the questions concerning their biological caregivers. Yet, the percentages of "yes" answers to the last two questions regarding their desire for permanency or adoption by their current caregivers convey a profound ambivalence that remains over time, with only half saying that they desire the current placement as a permanent home or desire adoption. However, the latent class and transition analysis reveal that considerable heterogeneity exists within this sample.

The identification of the three classes demonstrates three very different sets of perceptions about out-of-home care. The first class clearly wants to go home and does not perceive their placement as a place where they can stay until adulthood. Likewise, they do not desire to invest fully in this placement as a permanent home. The second class believes it is possible to live in the current placement until adulthood, but these youths are not yet sure that they can commit to this new environment.

This group remains hopeful about return to their biological families. The members of the third class are clear from the outset that they are not returning home and that they desire permanency in the current environment.

The transition analysis provides added detail. First, the majority of youths did not change their attitudes substantially within the 18-month time frame, particularly those whose feelings are extreme. Of those who are happy with their caregivers and clear that returning home is not something they either want or think is possible at time 1 (16%), none appear to move to other classes at time 2. This analysis would indicate that once this decision is made, youths embrace their new setting and see it as a permanent living situation.

Next, although youths whose perceptions change generally move toward more positive perceptions of care, characteristics of individuals and of the placement experience reveal more about what influences children's attitudes. First, as one would imagine, younger children appear more adaptable to new living situations than do older children. In this analysis, younger children were more likely than older children to remain content in care over two time points. This finding may be related to multiple developmental issues. Seven-to-ten-year-olds, the definition of younger children in this analysis, likely have not experienced the multiple physical and psychological transitions associated with adolescence. These changes may make accepting new adults as authority figures and mentors a more challenging proposition. Older children also may have created their own coping strategies for survival in their families of origin. Although functional in those environments, some of these strategies may be unacceptable in their new settings or impair their ability to form new relationships with both adults and peers. In these ways, older youths may be more vulnerable when entering placement.

Findings regarding placement stability indicate that although stability is certainly a goal to aspire to, placement change can be positive when a child is highly discontent in placement. In a seemingly counterintuitive finding, youths with higher levels of stability did not have positive attitude changes at time 2. However, this was true only for youths who were members of the highly discontent/want to go home class at time 1. The lack of placement change for this group, might have contributed to their negative attitudes toward care. In contrast, placement stability is associated with an ability to remain happy in care but still hopeful about going home. This finding is particularly important given that this class may represent a group of youths who are experiencing two potentially conflicting sets of attitudes. That is, these youths are happy in their current out-of-home placement yet maintain hope that they will return home. Theoretically, this scenario could be difficult if the child experienced these two feelings as competing loyalties. In reality, it is probably important for some sector of youths placed in out-of-home care to maintain both sets of attitudes when reunification is the goal of the child welfare system. The association between a stable placement that is viewed positively by the child and an ability to acknowledge an attachment to current caregivers and hope for reunification demonstrates the potential power of new relationships with caring adults to aid youths negotiating complex feelings and hopes toward their families of origin.

Mental health issues also have an association with one's level of contentment in care. Youths in the nonclinical range are less likely to make a radical transition from complete discontentment to being content in care and wanting to stay there. This may reflect that youths who do not evince mental health problems on entry into care are more able to accurately assess their feelings and are less likely to move between extremes of feeling. A nonclinical CBCL score at time 2 was associated with positive attitude changes, but not from one extreme category to another. Rather, a more step-by-step approach occurred for these youths, so that those who were discontent at time 1 were most likely to move to a position of happy but hopeful but were not yet fully committed to permanent placement at time 2. This is an area that warrants further investigation.

Limitations

In considering these results a few caveats should be considered. First, the available sample size contributed to a number of limitations. This analysis proposed a three-class solution to understanding adjustment to out-of-home care. However, there might be other trajectories that were not identified because the sample was not large enough to allow for estimation of a more complex model. Furthermore, the sample size limited the number of covariates that could be included in the model. Other characteristics, in addition to gender and mental health symptoms, may be associated with a young person's adjustment to out-of-home care. Specifically, relationship quality between the foster child and foster parent, previous patterns of attachment in a particular child, or school and neighborhood contexts may contribute to children's evolving views of placement.

Finally, this analysis centers on children who have been in care for a significant period of time. Many children stay in out-of-home placement for much shorter periods. Recent information from the Administration of Family and Children's Services indicates that 49% of youths who exited foster care in 2005 stayed 12 months or longer (U.S. Department of Health and Human Services, n.d.). However, the current analysis provides a useful starting point for understanding the differing paths by which youths adjust to out-of-home care.

Implications for Practice and Policy

The results of this analysis raise important issues for researchers, practitioners, and policymakers. First, the analysis points out the heterogeneity of perception and feeling within a group of young people involved in a similar difficult life circumstance--placement in out-of-home care. For researchers, this should encourage more investigations that "privilege the participant" by using analysis strategies that give voice to the variety of experiences that a large data set may represent.

For practitioners, these findings reiterate the importance of process in working with youths and the need to recognize that youths may be experiencing multiple feelings and hopes that sometimes appear contradictory. Furthermore, these perceptions of care can and do change over time, and some characteristics of the youths may influence those changing perceptions. For those working with youths in out-of-home placements because of child maltreatment concerns, it is important to assess these perceptions, recognize the complexity of youths' feelings about their current caregivers and their biological families, and to assist youths in sorting through their competing allegiances and ambivalent feelings. Furthermore, child welfare workers engaged with out-of-home caregivers may need to normalize the presence of these conflicting feelings in foster youths and assist caregivers in supporting youths as they work through these issues.

Finally, for policymakers, this analysis points out a need for flexibility in laws governing child placement. Many children want to be reunified with their biological families. Some may want to remain in their current out-of-home care arrangements but not legally break ties with their families of origin. Still others are clear that returning to their families of origin is the only satisfactory arrangement in their view. Of course, the child's wishes are not the only consideration in decisions about placement, and in many instances it may not be safe or realistic to act on a child's wishes. In constructing child welfare policy, the need for flexibility must be considered to allow frontline workers to truly act in a child's best interest. These youths hold many hopes and feelings about their past and future families. Part of attending to a child's well-being while in care means attending to these concerns.

Original manuscript received August 1, 2007

Final revision received February 22, 2008

Accepted May 9, 2008

REFERENCES

Achenbach, T. M. (1991). Manual for the Child Behavior Checklist 4-18 and 1991 profile. Burlington: University of Vermont.

Arbuckle, J. L. (1996). Full information estimation in the presence of incomplete data. In G.A. Marcoulides & R. E. Schumacker (Eds.), Advanced structural equation modeling: Issues and techniques (pp. 243-277). Hillsdale, NJ: Lawrence Erlbaum.

Barth, R. P. (1990). On their own: The experience of youth after foster care. Child &Adolescent Social Work Journal, 7, 419-440.

Berrick, J. D., Frasch, K., & Fox, A. (2000). Assessing children's experiences of out-of-home care: Methodological challenges and opportunities. Social Work Research, 24, 119-127.

Casey Family Programs. (2003, November). Assessing the effects of foster care: Early results from the Casey national alumni study. Seattle: Author.

Chapman, M.V., Wall, A. M., & Barth, R. P. (2004). Children's voices: Foster children's view of placement. American Journal of Orthopsychiatry, 74, 293-304.

Collins, L. M., Graham, J.W., Rousculp, S. S., & Hansen, W. B. (1997). Heavy caffeine use and the beginning of the substance use onset process: An illustration of latent transition analysis. In K. Bryant, M. Windle, & S. West (Eds.), The science of prevention: Methodological advances from alcohol and substance use research (pp. 79-99).Washington, DC: American Psychological Association.

Collins, L. M., & Wugalter, S. E. (1992). Latent class models for stage-sequential dynamic latent variables. Multivariate Behavioral Research, 27, 131-157.

Dowd, K., Kinsey, S., Wheeless, S., Suresh, R., & the NSCAW Research Group. (2004). National Survey of Child and Adolescent Well-being: Combined waves 1-4 data file user's manual. Research Triangle Park, NC: RTI International.

Fanshel, D., Finch, S.J., & Grundy, J. F. (1990). Foster children in a life course perspective. New York: Columbia University Press.

Fanshel, D., & Shinn, E. B. (1978). Children in foster care: A longitudinal investigation. New York: Columbia University Press.

Farmer, E.M.Z., Burns, B.J., Chapman, M.V., Phillips, S. D., Angold, A., & Costello, E.J. (2001). Use of mental health services by youth in contact with social services. Social Service Review, 75, 605-624.

Festinger, T. (1983). No one ever asked us: A postscript to foster care. New York: Columbia University Press.

Gibson, C. (2003). Privileging the participant: The importance of sub-group analysis in social welfare evaluations. American Journal of Evaluation, 24, 443-469.

Graham, J.W., Collins, L. M., Wugalter, S. E., Chung, N. K., & Hansen, W. B. (1991). Modeling transitions in latent stage-sequential processes: A substance use prevention example. Journal of Consulting and Clinical Psychology, 59, 48-57.

Haight, W.L., Black, J.E., Workman, C. L., & Tata, L. (2001). Parent-child interaction during foster care visits. Social Work, 46, 325-338.

Heinen, T. (1996). Latent class and discrete latent trait models: Similarities and differences. Thousand Oaks, CA: Sage Publications.

Horwitz, S. M., Balestracci, K.M.B., & Simms, M. D. (2001). Foster care placement improves children's functioning. Archives of Pediatrics and Adolescent Medicine, 155, 1255-1260.

James, S. (2004).Why do foster care placements disrupt? An investigation of reasons for placement change in foster care. Social Service Review, 78, 601-627.

Johnson, P. R., Yoken, C., & Voss, R. (1995). Family foster care placement: The child's perspective. Child Welfare, 74, 959-974.

Kufeldt, K., Armstrong, J., & Dorosh, M. (1995). How children in care view their own and their foster families: A research study. Child Welfare, 74, 695-715.

Lazarsfeld, P. F. (1950). The logical and mathematical foundation of latent structure analysis. In S.A. Stouffer, L. Guttman, E.A. Suchman, P. E Lazarsfeld, S.A. Star, & J. A. Clausen (Eds.), Measurement and prediction (pp. 362-412). Princeton, NJ: Princeton University Press.

Leitenberg, H., Burchard, J. D., Healy, D., & Fuller, E.J. (1981). Non-delinquent children in state custody: Does type of placement matter? American Journal of Community Psychology, 9, 347-360.

Muthen, B. O., & Muthen, L. K. (1998-2006). Mplus user's guide (4th ed.). Los Angeles: Author.

Peters, J. (2005).True ambivalence: Child welfare workers' thoughts, feelings, and beliefs about kinship foster care. Children and Youth Services Review, 27, 292-314.

Pilowsky, D.J., & Kates, W. G. (1996). Foster children in acute crisis: Assessing critical aspects of attachment. Journal of the American Academy of Child and Adolescent Psychiatry, 35, 1095-1097.

Taussig, H. N., Clyman, R. B., Landsverk, J. (2001). Children who return home from foster care: A 6-year prospective study of behavioral health outcomes in adolescence. Pediatrics, 108, E10.

U.S. Department of Health and Human Services, Administration for Children and Families, Administration on Children, Youth and Families, Children's Bureau. (n.d.). Preliminary estimates for FY 2005 as of September 2006. Retreived February 14, 2008, from www.acf.hhs.gov/programs/cb

Wald, M. S., Carlsmith, J. M., & Leiderman, P. H. (1988). Protecting abused and neglected children. Stanford, CA: Stanford University Press.

Wedeven, T., Pecora, P.J., Hurwitz, M., Howell, R., & Newell, D. (1997). Examining the perceptions of alumni of long-term family foster care: A follow-up study. Community Alternatives: International Journal of Family Care, 9, 88-106.

Wilson, U, & Conroy, J. (1999). Satisfaction of children in out-of-home care. Child Welfare, 78, 53-68.

Mimi V. Chapman, PhD, is associate professor, School of Social Work, University of North Carolina at Chapel Hill, 325 Pittsboro Street, CB 3550, Chapel Hill, NC 27599-3550; e-mail: mimi@email.unc.edu. Sharon L. Christ, MA, is a doctoral student in sociology and a research associate, Odum Institute, University of North Carolina at Chapel Hill. An earlier version of this article was presented at the Society for Social Work Research, January 2006, San Antonio, TX.
Table 1: Population Distribution of Indicator Variables of the Latent.
Classes for Children in Out-of-Home Care (unweighted n = 290)

 Baseline 18 Months
Latent Class Variable Indicators % Yes % Yes

Can you keep living here until you grow up? 63.85 7-.()2
Do you think you'll be living with your
 [current caregiver] next year? 56.27 76.49
Do you think you will live with your real
 mom or real dad again? 60.13 38.28
If you went back to living with your real
 mom or real dad, would things be different
 than they were before? 63.09 59.52
Do you want this to be your permanent home? 41.11 53.55
Do you want your [current caregiver]
 to adopt you? 32.18 48.21

Table 2: Model Fit Statistics for Two,
Three, Four, and Five Class Models at
Baseline and 18 Months Post-Baseline

 Adjusted
 AIC BIC BIC Entropy

 Baseline

2 classes 1836.15 1891.20 1843.63 0.828
3 classes 1745.51 1833.58 1757.47 0.901
4 classes 1675.79 1796.89 1692.24 0.936
5 classes 1644.01 1798.15 1664.96 0.902

 18 Months

2 classes 1625.57 1680.62 1633.05 0.877
3 classes 1579.93 1668.01 1591.90 0.874
4 classes 1543.83 1664.94 1560.29 0.844
5 classes 1530.70 1684.83 1551.64 0.872

Notes: AIC = Akaike information criterion; BIC = Bayesian
Information criterion. The boldface data represent the
chosen three-class model, which was the most stable at
baseline and at 18 months.

Table 3: Latent Class Profiles from the Transition Model-for the
Population of Children in Out-of-Home Care

 Want to Happy but Content
 Go Home Hopeful in Care
Latent Class Variable Indicators % Yes % Yes % Yes

Can you keep living here until
 you grow up? 45.4 99.6 88.70
Do you think you'll be living with
 your [current caregiver] next year? 42.6 96.4 76.80
Do you think you will live with
 your real mom or real dad again? 78.4 38.6 0 (a)
If you went back to living with
 your real mom or real dad, would
 things be different than they
 were before? 82.9 70.4 0 (a)
Do you want this to be your
 permanent home? 2.3 95.1 85.5
Do you want your [current caregiver]
 to adopt you? 7.2 63.7 84.5

(a) These values were fixed because of empty cells.

Table 4: Latent Class Transition
Probabilities for Children in
Out-of-Home Care

 Want to Happy but Content
 Go Home Hopeful in Care

Want to go home 0.629 0.285 0.087
Happy but hopeful 0.196 0.554 0.249
Content in care (a) 0 0 1

(a) Transition probabilities were fined because of empty cells.

Table 5: Latent Status Possibilities:
Percentage of Total Sample of
Children in Out-of-Home Care
Likely to Be in Each Status

 % n

Negative nonmover 36 105
Hopeful nonmover 14 41
Content nonmover 16 46
Hopeful to contentment 7 19
Moving toward discontentment 5 15
Previously discontent to content 5 15
Moving toward hopeful from discontent 17 49
Once content, now discontent --
Once content, now moving toward home --

Note: Dashs indicate that transition does not occur.

Table 6: Regression Results for the Probability of Transition
Status Membership, by Child Characteristics

 Gender Age

 B p B P

Want to go home to
 happy & hopeful 0.14 .01
Want to go home to content 0.09 .03
Remains Want to go home
Happy & hopeful to content 0.15 .003
Remaining happy & hopeful 0.14 .016
Remaining content -0.25 .0004 0.18 .046

 Mental Placement
 Health Stability

 B P B P

Want to go home to
 happy & hopeful 0.14 .025 -0.07 .013
Want to go home to content
Remains Want to go home -0.23 .019
Happy & hopeful to content
Remaining happy & hopeful 0.13 .0142
Remaining content

Note: Only statistically significant findings at alpha
[less than or equal to] .05 are noted.
COPYRIGHT 2008 National Association of Social Workers
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2008 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Chapman, Mimi V.; Christ, Sharon L.
Publication:Social Work Research
Article Type:Report
Geographic Code:1USA
Date:Sep 1, 2008
Words:7087
Previous Article:Accomplishments and future directions for social work research: reflections as editor-in-chief.
Next Article:Pathways to drug and sexual risk behaviors among detained adolescents.
Topics:

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters