The first four months in a new foster placement: psychosocial adjustment, parental contact and placement disruption.

Delfabbro, Paul H.

Intake and four-month follow-up measures were obtained for 235 children referred into a new foster care placement over a 12-month period in the Australian State of South Australia. Twenty-five percent of the sample returned home within 4-months, and for those who remained in care throughout, there had been modest gains in behavior, psychological adjustment and adjustment at school. On the other hand, there were considerable levels of placement disruption, a high degree of non-compliance with parental visiting plans, and a high proportion of children fell outside ninety-five percent confidence intervals for the general adolescent population on most well-being measures, particularly conduct disorder.

Introduction

Despite the fact that child welfare legislation everywhere advances child well-being as one of its most fundamental objectives, efforts to measure the well-being of children in state care have been surprisingly rare and unsustained. Altshuler and Gleeson (1999), for example, recently noted that measures of success in foster care are dominated by indicators of permanency and safety, while child well-being is rarely incorporated into administrative databases or built into the evaluation of system performance. No doubt one of the reasons for this omission is that whereas permanency and safety can be readily inferred from administrative data such as re-abuse and re-referral rates, the measurement of child well-being is a more subjective and potentially labour-intensive task.

In a recent paper on the measurement of child well-being, Barber and Delfabbro (2000) argued that for well-being assessments to become routine, there is a need for briefer, more useable measures than are currently available; measures that can be incorporated into the day-to-day casework of child welfare professionals. Many of the more commonly advocated measures of child well-being such as the Child Well-Being Scales (Magura & Moses, 1986) and the Child Behavior Checklist (Achenbach, 1981) are much too laborious to pass this test. Another problem with the available research into foster child well-being is that most studies have been cross-sectional (see Altshuler & Gleeson, 1999 for a review). In the most common research design, the functioning of children in foster care is compared with that of children in the general population or from comparable groups in the child welfare population at a single point in time (cf. Kinard, 1994). Such designs provide no adequate baseline against which to compare change in foster care outcomes. What longitudinal studies have been conducted are mainly retrospective. Large archival data-sets, such as those routinely maintained by agencies, have been used to examine the long-term outcomes of care (e.g., Courtney, 1994, 1995; Courtney & Wong, 1996; Fernandez, 1998; Goerge, 1990). These studies have proved highly effective in predicting changes in case-status over time, but have been limited by the range of variables included, the sophistication of the measures available, and by the absence of follow-up measures more proximal to the outcomes predicted. For example, it is questionable whether particular outcomes can be clearly associated with factors such as abuse which may have occurred 5-10 years earlier.

Accordingly, prospective longitudinal studies are increasingly advocated in the child welfare field (Courtney et al., 1998; Fanshel, 1975a; Wulczyn, 1997). In addition to being able to compare subsequent results with a consistent baseline, prospective studies are in a position to collect a greater volume of information, and to choose what information should be collected. Archival or case-file information can be combined with child self-reports and reports from others who have regular face-to-face contact with the child. Furthermore, although concerns can be raised about potential biases resulting from the selective loss of subjects over time, a prospective study often has the capacity to identify, and maybe control for, any systematic differences between the retained sample and those who drop out.

The present study represents the first stage of a prospective study into foster child well-being. Brief measures of behavioral disorder, psychological adjustment and adjustment at school were routinely administered at intake and again four months later. The frequency and reliability of parental visiting was also assessed because it has been consistently related not only to family reunification (Fanshel, 1975b; Gibson et al., 1984; Lawder et al., 1986; Milner, 1987; Seaberg & Tolley, 1986) but also to children's well-being in care (Cantos, Gries, & Slis, 1997; Fanshel & Shinn, 1978; Poulin, 1985). Given that details of family contact are often not adequately documented in case-files (Grigsby, 1994; Hess, 1982; Hess, Mintun, Moelhman, & Pitts, 1992), the use of face-to-face interviews in the present study, combined with pre- and post analysis, may offer a more accurate and detailed insight into this key area of practice.

Method

Study Design

Two-hundred-and-thirty-five children referred into out-of-home care over a twelve-month period were recruited for the study. Baseline information was obtained from the children's case files and from face-to-face interviews with their social workers. Four-months later, the 164 (70%) children who remained in care were followed-up, (1) and their social workers reinterviewed. This study reports on placement movements between intake and follow-up for the whole sample (n = 235), and on the behavioral, psychological, and educational progress of the 170 children who remained in care throughout the study period. Scores on behavior and psychological measures for adolescents in the sample were also compared with a normative sample (n = 985) of adolescents from the general population.

The Foster Care Sample

The original sample involved 235 children (121 boys, 114 girls) with a mean age of 10.8 years and an age range of 4-17 years. Children were selected if they were referred for a new placement between May 1998 and April 1999. Excluded from the sample were children on detention orders, children placed into supported accommodation, those referred for family preservation services, those referred for respite from a continuing placement, and those with placements expected to last for less than 2 weeks. The final sample represents the entire cohort of children meeting the selection criteria referred via the central referral agency for both metropolitan and rural areas of South Australia during the study period. Included in the sample were 39 (16%) Aboriginal children and 195 (84%) non-Aboriginal children (1 missing item). Sixty-three (27%) children were from rural areas of South Australia and 172 (73%) were from the metropolitan area of Adelaide. A breakdown of the intake sample by age showed that 65 (28%) were aged 4-8 years, 80 (34%) were 9-12 years and 90 (38%) were aged 13-17 years. At intake there were 110 children aged between 12 and 17 years but by the follow-up point 85 adolescents in that age group remained in the sample.

The Normative Sample

Comparative data for the adolescent foster children were obtained from a data-set involving 985 parents drawn form the general population in Canada. All parents in the sample had children drawn from 95 Canadian schools in the province of Alberta. One school was randomly selected from each of the 9 school districts and within these schools individual students were selected by stratified random sampling according to age group from 12 through 17 years. Individual school districts were responsible for drawing the sample based on a required sample size. Following identification of the students, mailing labels containing parents' names and addresses were prepared by each school district. Each of these parents was dispatched a letter describing the study and inviting them (the parent) to participate. The parents who agreed to participate had children whose mean age was 14.31 years (s.d. = 2.51), comprising 502 males (51%) and 483 females (49%). Although the normative sample and the foster care sample came from different countries, recent research indicates that the general Canadian and Australian adolescent populations display very similar levels of well-being (Barber, 2001).

Measures and Procedure

1. The Foster Care Sample

In order to recruit the foster care sample, referral records at the central agency were monitored each week. Data on the children selected were recorded along with the contact details and location of the social worker responsible for each case. Information was collected from central agency records and government databases, and verified in face-to-face interviews with social workers. Four-months later most of the measures described below were reobtained for those children (n = 164) who remained in care.

Among the variables extracted from agency records were: (a) demographic characteristics, (b) placement history, and (c) details of current placement, including type of legal order and type and frequency of contact with family of origin. The types of contact included: information only, indirect contact (e.g., telephone), direct visits, and overnight stays. Frequency of each type was measured on a 6 point scale: 1 = never, 2 = monthly or less often, 3 = 2-3 times per month, 4 = once per week, 5 = 2-6 times per week, 7 = daily or more often. The referral form also provided a checklist of factors that were taken into account when making the referral, such as whether or not the child had been the victim of abuse, child mental health problems, suicide attempts, substance abuse, developmental delay, self-destructive behaviors, offending, truancy, and other general behavioral problems.

Among the measures obtained from interviews with the child's social worker was an abbreviated form of Boyle et al.'s (1987) Child Behavior Checklist (CBC). The CBC contains four subscales: conduct disorder, hyperactivity, somatization disorder and emotional disorder. All subscale items were designed by Boyle et al (1987) to operationalize DSM criteria for the relevant disorder. The items selected from the CBC for inclusion in this study were those found in a previous study involving the senior author of over 2,000 junior and senior high school students to possess the highest item-total correlations within their relevant subscales (Barber, Bolitho & Bertrand, 1998; 1999a; 1999b). Using this approach, 6 of the 15 conduct disorder items could be extracted without negatively affecting alpha for the conduct disorder subscale. Similarly, 3 of the 6 items comprising Boyle et al's (1987) hyperactivity subscale, 5 of their 13 emotionality items, and 4 of their 11 somatization items were selected for the remainder of the abbreviated CBC. Internal consistency for each of the abbreviated subscales was found to be acceptable using the intake sample from the present study: [alpha] = .83, .87, .82, and .84 for the conduct disorder, hyperactivity, emotional disorder, and somatization disorder subscales respectively.

In addition to behavioral and psychological adjustment measures, social workers were asked to respond to 10 items concerning the child's behavior and adjustment at school. All items were scored on 4-point scales: 1 = often, 2 = sometimes, 3 = rarely, 4 = never. Four items referred to general school performance: `Has been well-organised', `Has been interested in his/her studies', `Has produced work of a good standard', `Has been attentive in class'. Six items referred to the child's behavior and level of cooperation: `Has not completed homework or set work', `Has been disruptive in class', `Has refused to take part in school activities', `Has been disciplined by teachers and other staff members', `Has been late to class' and `Has wagged (skipped) school'. If the child was not attending school, workers were asked to indicate why this was so on a short checklist which included: left-school, excluded, school refusal. Finally, as an objective indicator of school adjustment, social workers were asked to indicate how many times the child had changed school, been suspended or excluded during the 12 months prior to entering the study (at intake) and in the four-months since intake (at follow-up).

2. The Normative Sample

Parents of the adolescent children selected for the sample were asked to complete the long form of Boyle et al.'s (1987) CBC in relation to their child. Items included in the abbreviated CBC were extracted from the parent dataset for analysis in this paper.

Results

1. Placement Change and Stability

The overall sample can be divided into three groups: 1) new referrals to care (n = 40); 2) those who were returning to care after going home from a previous placement (n = 89); and 3) those who were currently in care but were changing placements (n = 106). Figure 1 identifies these 3 groups and describes the status of the sample at the follow-up point.

[FIGURE 1 OMITTED]

Analysis of the reasons for the 123 placement changes during the four-month period revealed that 83 (67%) of the children were moved for reasons beyond their control, such as because better or more permanent arrangements became available or because of changes in carer circumstances. However, 40 (33%) of the children who experienced at least one placement change were moved because the foster carer had been unable to cope with the child's behavior. The mean number of placements experienced by these 40 children in the 4-month period was 5.7 (s.d. = 4.2).

Table 1 shows that placement status at follow-up varied significantly depending on the group to which the children belonged at intake. Proportion difference tests showed that new referrals were significantly more likely to return home than the other two groups, z = 3.99, p < .01, whereas children who were already in the system at intake were no more likely to change placement overall, z = 1.61, p > .05, but were more likely to experience a placement change due to their disruptive behavior, z = 3.25, p < .01. As 59 of the original sample had returned home by the follow-up point (and were therefore no longer under State supervision), comparisons between intake and follow-up measures in the remaining sections were confined to the children who remained in care throughout the study period (n = 164).

2. Family Contact

Analysis of case plans indicated that almost all (n = 212 or 90%) of the children in this study did have explicit plans in relation to family contact, although in 52 cases (22%), the plan was for no contact (2) and only 127 (54%) were expected to have direct, personal contact, as opposed to telephone contact. When contact was planned, in whatever form, it was usually planned to occur at least weekly. Furthermore, half of the children for whom no contact was planned did, in fact, receive family contact over the period anyway. McNemar change tests, which compared the intake and follow-up frequencies for each contact category, showed that children were significantly less likely to have `no contact' than had been anticipated (predicted=22%, actual=11%, p < .05), and stayed overnight more often than predicted (intake=11%, actual=27%, p < .001). Further analyses compared the predicted and actual frequency of contact by collapsing the original 6 response categories into separate binary frequencies (1 month or less vs. 2-3 times per month or more). This analysis showed that frequency of all three types of contact (indirect, direct visits and overnight stays) did not differ from what had been predicted.

Further analyses revealed that the generally high level of conformity with family contact plans at the aggregate level masks a considerable degree of non-compliance at the individual level. Planned contact did not occur between 6% and 19% of the time, depending on the type of contact in question. And in around 20% of cases, contacts were made when no such plans had been made.

3. Psychological Adjustment

Table 2 presents mean item scores on three of the CBC subscales for the children in care at intake and follow-up. (There were no changes in somatization scores). Mean item scores were obtained by adding all items completed and dividing by the total number of items completed. Within samples analysis of mean item conduct score revealed an improvement in conduct from intake to follow-up. Analysis of individual conduct subscale items showed that the change in overall score was due to a significant reduction in 2 of the 6 conduct items, with children being less likely to destroy property or lie and cheat once they had been in care for 4 months. Similarly, there was a significant reduction in overall hyperactivity, due in this case to 2 of the 3 items: concentration problems and distractibility. Finally, the change in overall emotionality score was due to improvement in 1 of the 5 items--"worried a lot". Taken together, then these analyses suggest that children in care were generally better behaved, less agitated and less worried than they had been at intake. However, assessments of effect size based upon a comparison of mean difference scores and the standard deviations (Cohen, 1992), revealed that the effect sizes (d) for all comparisons were only small (i.e., < 0.30).

Figure 2 depicts the distribution of mean conduct scores for the normative sample and for children between the ages of 12 and 17 years who were in foster care at intake and follow-up. Sixty-percent of foster children at intake and 41% at follow-up fell outside the 95th percentile for the normative population. Notwithstanding the apparent decline in the number of foster children within this outlier group, independent samples t-tests showed that the foster care sample scored significantly higher than the normative sample on overall mean score at both points in time ([p.sup.s] < .001). Similar results were obtained for the other two CBC subscales. Twenty-eight percent of foster children fell outside the 95th percentile for hyperactivity at intake, compared with 21% at follow-up, and 25% fell outside the 95th percentile for emotionality at intake compared with 23% at follow-up.

[FIGURE 2 OMITTED]

4. Behavior at School

Table 3 indicates that a substantial number of children were experiencing significant problems in school at intake. Approximately a quarter were often disruptive in class and were not completing set work, over a third were often disciplined by staff members, and nearly a third were skipping school often or sometimes. Between 1 in 5 and 1 in 3 children at intake were reported as rarely or never being attentive in class, interested in their studies, or producing work of a good standard. Encouragingly, results showed some improvement in school performance. McNemar change tests compared the relative percentage of children in the often + sometimes group for each item compared with the percentage in the rarely + never category. Children were significantly less likely to skip school in their new placement (29% at intake versus 16% at follow-up, p < .05), and to refuse to take part in school activities (51% at intake versus 35% at follow-up, p < .05). On the other hand, social workers reported no significant improvement in how much interest children were showing in their studies, in the quality of work produced, level of organisation, or general attentiveness.

A more objective measure of school adjustment could be derived by comparing the rate of school suspensions and exclusions prior to and following placement. Sixty of the children in receipt of care throughout the study period had been suspended at least once in the 12-months prior to the study for a mean duration of 14 days, and 9 had been excluded. During the first follow-up period, 26 children were suspended with 8 having been suspended on 3 or more occasions. Each suspension was for approximately 2-7 days. There were also 6 exclusions, with 4 of these children not returning to school at all. An approximate suspension rate can be calculated by dividing the pre-placement mean by 3 to give mean suspension per quarter rate (i.e., 0.95/4 =0.32). This compares with 0.17 for the first follow-up period using the same algorithm. Thus, both social worker ratings and suspensions data demonstrate a decrease in problematic school behaviors during the first four months in a new placement.

Discussion

This paper began by arguing the need for quick, reliable measures of child well-being that can be readily integrated into the casework of foster care workers. While the abbreviated CBC scales developed in the present study certainly proved to be both easy to administer and reliable, a feature of the dataset was the large amount of missing data on the abbreviated CBC because social workers simply did not know the answer. This finding underscores an observation we have made more than once before (Barber & Delfabbro, 2000; Delfabbro, Barber & Cooper, 2002) that the amount and quality of information on which child welfare decisions are commonly based tends to be very limited. We have further suggested that this problem is endemic in the child welfare field because so many of the relevant variables refer to inherently private states and behaviors and because respondents frequently have some incentive to disguise the truth. Interestingly, the degree of missing data in our abbreviated CBC corresponds quite closely to that recorded by Boyle et al. (1987) for the longer version of the CBC.

Notwithstanding the problem of missing data, results of this prospective study are consistent with the conclusion that foster placement tends to be accompanied by improvements in the short-term in levels of conduct disorder, hyperactivity and emotional disorder. The present study also found a statistically significant improvement in foster children's attendance and participation at school between intake and follow-up. This improvement in school behavior was reflected in a lower rate of school exclusions once the children came into foster care.

Offsetting these positive results somewhat was the finding that when the well-being of adolescents in foster care was compared with a normative sample from the general population, a sizeable proportion of the foster children fell outside acceptable confidence intervals at intake and at follow-up. It must be acknowledged, however, that the extent to which this result is attributable to the different raters used for the two groups of children (social workers for foster children versus parents for normative children) is unknown. It must also be emphasised that in the absence of an adequate control condition, improvements in behavior and well-being cannot be attributed to foster placement itself. Among the more obvious threats to the internal validity of this conclusion, for example, are: (a) child maturation and (b) instrumentation, particularly changes in social worker expectations. Moreover, as Nelson, Singer, and Johnson (1978) note, improvements in the child's functioning on entering care may be short-lived as most children can be expected to conform at least temporarily to the behavioral expectations of a new setting. In order to assess this possibility, further follow-up of the children in this study is planned.

Results of the family reunification data showed that nearly twenty-five percent of the sample had returned home prior to the four-month follow-up point, although this aggregate figure masked considerable variation between referral types. Of particular concern is that the most difficult group to reunify--those referred for a change of placement--was also the most numerous. Whereas around forty percent of new referrals and one-third of children returning to care from home were reunited with their families within four months, only around nine percent of those referred for a change of placement could be reunified within four-months. Moreover, around half of the children referred for a change of placement were forced to change placement again at least once within four-months, in almost fifty-percent of these cases because the foster carer was unwilling or unable to tolerate the child's disruptive behavior any longer. Results of the family contact data also suggest that social workers were aware of the importance of explicit plans in relation to family contact, as ninety-percent of the children in care throughout the study period did have a family contact plan. However, the data showed a considerable level of non-compliance with the plan.

Taken together, then, our findings present a mixed picture of the first four months in foster care. Of particular concern is the number of children who appear to be adrift in the foster care system. That is to say, they wander from foster home to foster home, neither returning to their families nor settling down with alternative carers. The present study suggests that over half of the children in South Australia who are referred into out-of-home care will, for one reason or another, be forced to change placement at least once in the first four-months. While it is acknowledged that some of these moves constitute transition arrangements, such as where a child needs an emergency placement while more stable arrangements are worked out, this does not alter the fact that placement change is a very disruptive experience. For this reason, one of the most fundamental objectives of an alternative care system must surely be to provide placement stability from the outset.

Table 1

Status at follow-up by origin of child

 New into Returning Changing
 Care to care placement
 (n = 40) (n = 89) (n = 106)

Returned home 20 (50.0%) 29 (32.6%) 10 (9.4%)
Still in care
In 1 stable placement 11 (27.5%) 30 (33.7%) 35 (33.0%)
Changed placement 7 (17.5%) 30 (33.7%) 54 (50.9%)
Change due to child's 3 (7.5%) 10 (11.2%) 27 (25.5%)
 behavior

Table 2

Means, standard deviations (in parentheses), and valid cases for CBC
subscales at intake and at 4 months

 Intake 4-months t-value Effect
 size(d)

Conduct
Damaged property 0.50 (0.64) 0.40 (0.67) 1.88 0.15
Destroyed property 0.68 (0.67) 0.48 (0.72) 3.01 ** 0.29
Disobedient at school 0.92 (0.75) 0.84 (0.76) 1.07 0.11
Lied or cheated 1.13 (0.69) 0.97 (0.76) 2.41 * 0.22
Stole things 0.49 (0.70) 0.40 (0.66) 1.53 0.13
Physical attacks 0.52 (0.69) 0.41 (0.63) 1.94 0.17
Mean score 0.71 (0.51) 0.59 (0.52) 3.02 ** 0.21

Hyperactivity
Couldn't concentrate 1.33 (0.72) 1.17 (0.81) 2.26 * 0.21
Couldn't sit still 1.00 (0.79) 0.93 (0.82) 1.21 0.09
Distractible 1.20 (0.72) 1.02 (0.79) 2.89 ** 0.24
 Mean score 1.17 (0.66) 1.00 (0.70) 3.08 ** 0.28

Emotionality
Unhappy, sad or 1.21 (0.59) 1.14 (0.61) 1.14 0.12
 depressed
Not as happy as other 1.21 (0.60) 1.11 (0.69) 1.43 0.16
 children
Nervous and tense 0.94 (0.76) 0.86 (0.75) 1.14 0.11
Too fearful or anxious 0.81 (0.73) 0.78 (0.64) <1 0.04
Worried a lot 1.11 (0.65) 0.95 (0.72) 2.22 * 0.23
 Mean score 1.07 (0.53) 0.93 (0.51) 2.80 * 0.19

* v < .05. ** v < .01

Table 3

School performance and adjustment at intake and follow-up (follow-up
figures in bold, negative items shaded)

 n Often Sometimes

Has not completed 21 (25.6) 35 (42.7)
 homework or set 82 16 (19.5) 34 (41.5)
 work.
Has been attentive in 29 (30.9) 48 (51.1)
 class. 94 33 (35.1) 38 (40.4)
Has been disruptive in 27 (22.9) 41 (34.7)
 class. 118 19 (16.1) 50 (42.4)
Has refused to take 11 (13.4) 31 (37.8)
 part in activities. 82 9 (11.0) 20 (24.4)
Has been well- 21 (26.9) 20 (25.6)
 organized. 78 17 (21.8) 29 (37.2)
* Has been interested 26 (29.2) 43 (48.3)
 in studies. 89 27 (30.3) 43 (48.3)
Has been disciplined 31 (33.7) 41 (44.6)
 by staff. 92 23 (25.0) 38 (41.3)
Has produced work of 18 (21.2) 43 (50.6)
 a good standard. 85 21 (24.7) 38 (44.7)
Has been late to class. 18 (22.5) 15 (17.6)
 80 7 (8.8) 14 (16.5)
Has wagged 16 (17.0) 11 (11.7)
 (skipped) school. 94 6 (6.4) 9 (9.6)

 Rarely Never

Has not completed 16 (19.5) 10 (12.2)
 homework or set 9 (11.0) 13 (15.9)
 work.
Has been attentive in 16 (17.0) 1 (1.2)
 class. 12 (12.7) 3 (3.6)
Has been disruptive in 14 (11.9) 16 (13.6)
 class. 19 (16.1) 20 (16.9)
Has refused to take 16 (19.5) 24 (29.3)
 part in activities. 13 (15.6) 33 (40.2)
Has been well- 28 (35.9) 9 (11.5)
 organized. 19 (24.4) 6 (7.7)
* Has been interested 16 (18.0) 4 (4.5)
 in studies. 11 (12.4) 3 (3.4)
Has been disciplined 9 (9.8) 11 (11.6)
 by staff. 10 (10.9) 8 (8.7)
Has produced work of 18 (21.2) 6 (7.1)
 a good standard. 11 (12.9) 6 (7.1)
Has been late to class. 9 (11.3) 38 (47.6)
 10 (12.5) 33 (41.3)
Has wagged 6 (6.4) 61 (64.9)
 (skipped) school. 10 (10.6) 62 (66.0)

Notes

(1.) Follow-up was undertaken for children whose case were still open (or only recently closed) irrespective of their status. However, our analyses are confined to those children still in care.

(2.) Contact is often proscribed because of the risk of abuse to the child.

References

Achenbach, T.M. (1981) Child Behavior Checklist for Ages 4-16. Brlington, VT: University Associates in Psychiatry.

Altshuler, S.J. & Gleeson, J.P. (1999) Completing the evaluation triangle for the next century: Measuring child "well-being" in family foster care. Child Welfare, 78, 125-147.

Barber, J.G. (2001) Relative misery and youth suicide. Australian & New Zealand Journal of Psychiatry, 35(1), 49-57.

Barber, J.G., Bolitho, F. & Bertrand, L. (1998) Age and gender differences in the predictors of adolescent drinking. Social Work Research, 22, 164-172.

Barber, J.G., Bolitho, F. & Bertrand, L. (1999a) The predictors of adolescent smoking. Journal of Social Services Research, 26, 51-66.

Barber, J.G., Bolitho, F. & Bertrand, L. (1999b) Intrapersonal versus peer group predictors of adolescent drug use. Children and Youth Services Review, 21, 565-579.

Barber, J.G. & Delfabbro, P.H. (2000) A strategy for the routine assessment of child well-being by child welfare agencies. The Journal of Social Work Research and Evaluation.

Barth, R.P. & Berry, M. (1987) Outcomes of child welfare services under permanency planning. Social Service Review, 61, 71-90.

Boyle, M.H., Offord, D.T., Hofman, H.G., Catlin, G.P., Byles, J.A., Cadman, Crawford, J.W., Links, ES., Rae-Grant, N.I. & Szatmari, P. (1987) Ontario Child Health Study: I. Methodology. Archives of General Psychiatry, 44, 826-831.

Bryce, M.E. & Ehlert, R.C. (1971) 144 foster children. Child Welfare, 50, 499-503.

Cantos, A.L., Gries, L.T., & Slis, V. (1997) Behavioral correlates of parental visiting during family foster care. Child Welfare, 76, 309-329.

Claburn, W.E., Magura, S. & Resnick, W. (1976) Administrative case review for foster care: A brief national assessment. Child Welfare, 55, 395-405.

Cohen, J. (1992) A power primer. Psychological Bulletin, 112, 155-159.

Courtney, M.E. (1994). Factors associated with the reunification of foster children with their families. Social Service Review, 68, 81-108.

Courtney, M.E. (1995). Reentry to foster care of children returned to their families. Social Service Review, 69, 226-241.

Courtney, M.E. & Wong, Y.I. (1996). Comparing the timing of exits from substitute care. Children and Youth Services Review, 18, 307-334.

Courtney, M.E., Piliavin, Grogan-Kaylor, A., & Nesmith, A. (1998). Foster youth transitions to adulthood: a longitudinal view of youth leaving care. School of Social Work and Institute for Research on Poverty, University of Wisconsin: Madison.

Delfabbro, P.H., Barber, J.G. & Cooper, L.L. (2002) Longitudinal foster care research: The issue of data quality and reliability in caseworker reports. Submitted.

Fanshel, D. (1975a). Status changes of children in foster care: final results of the Columbia University longitudinal study. Child Welfare, 55, 143-177.

Fanshel, D. (1975b) Parental visiting of foster children in foster care: Key to discharge? Social Service Review, 49, 493-514.

Fanshel, D. & Shinn, E. (1978) Children in Foster Care: A Longitudinal Investigation. New York: columbia University Press.

Fernandez, E. (1999). Pathways in substitute care: representation of placement careers of children using event history analysis. Children and Youth Services Review, 21, 177-216.

Gibson, T.L., Tracy, G.S., & DeBord, M.S. (1984) An analysis of the variables affecting length of stay in foster care. Children and Youth Services Review, 6, 135-145.

Goerge, R.M. (1990). The reunification process in substitute care. Social Service Review, 64, 422-457.

Grigsby, R.K. (1994) Maintaining attachment relationships among children in foster care. Families in Society, 75, 269-276.

Hess, P. (1982) Parent-child attachment concept: Crucial for permanency planning. Social Caseowrk, 63, 46-53.

Hess, P., Mintun, G., Moelhman, A., & Pitts, G. (1992) The Family Connection Center: An innovative visiting program. Child Welfare, 71, 77-88.

Katz, L. (1990) Effective permanency planning for children in foster care. Social Work, 35, 220-226.

Kinard, E.M. (1994). Methodological issues and practical problems in conducting research on maltreated children. Child Abuse and Neglect, 18, 645-656.

Lawder, E., Poulin, J.E., & Andrews, R. (1986) A study of 185 foster children five years after placement. Child Welfare, 65, 241-245.

Magura, S. & Moses, B.S. (1986) Outcome Measures for Child Welfare Services: Theory and Applications. Washington, DC: Child Welfare League of America.

Maluccio, A.N., Fein, E. & Olmstead, K.A. (1986) Permanency Planning for Children: Concepts and Methods. New York: Tavistock Publications.

Milner, J.L. (1987) An ecological perspective on duration of foster care. Child Welfare. 66, 113-123

Nelson, R.H., Singer, M.J., & Johnsen, L.O. (1973) The application of a residential treatment model. Child Care Quarterly, 7, 164-175.

Poulin, J. (1985) Long term foster care, natural family attachment and loyalty conflict. Journal of Social Service Research, 9, 17-29.

Seaberg & Tolley (1986) Predictors of the length of stay in foster care. Social Work Research and Abstracts, 22, 11-17

Wulczyn (1997). Foster care outcomes and performance standards: a report of the task force on foster care performance standards. Chapin Hall Center for Children: New York.

JAMES G. BARBER
University of Toronto
Faculty of Social Work

PAUL H. DELFABBRO
University of Adelaide
Department of Psychology