• No results found

Foster care placement instability : a systematic review and meta-analysis

N/A
N/A
Protected

Academic year: 2021

Share "Foster care placement instability : a systematic review and meta-analysis"

Copied!
41
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

Foster care placement instability

A systematic review and meta-analysis

Masterscriptie Forensische Orthopedagogiek Graduate School of Child Development and Education Universiteit van Amsterdam J. Baart & S. Admiraal 10472770 & 10457461 Begeleiding: Geert Jan Stams & Mark Assink Amsterdam, Juli 2016

(2)

2 Contents Abstract ... 3 1. Introduction ... 4 2. Method ... 7 2.1 Sample of studies ... 7 2.2 Publication bias ... 8 2.3 Coding of studies ... 8 2.4 Statistical analyses ... 9 3. Results ... 11

3.1. Descriptives, central tendency and variability, and assessment of missing data ... 11

3.2. Moderator analyses ... 12

3.2.1. Behavioral problems of the child ... 13

3.2.2. Low quality of foster parenting ... 13

3.2.3. Non-kinship care ... 13

3.2.4. Older age ... 13

3.2.5. Maltreatment of the child ... 14

3.2.6. Ethnic minorities ... 14

3.2.7. Child’s previous number of out-of-home placements ... 14

4. Discussion ... 211

4.1 Overall and moderator effects of domains of risk factors ... 211

4.2 Limitations ... 255

4.3 Conclusion ... 277

References ... 288

(3)

3

Abstract

This review examined risk factors for placement instability of children in foster families. A series of multilevel meta-analyses were performed to examine the effects of several risk factors for instability of foster care placement. We included 38 studies reporting on 182 effect sizes for risks associated with placement instability. The largest significant effects were found for behavioral problems of the child (r = .33, p < .001), although this effect was substantially reduced after correction for publication bias (r = .09, p < .05), and low quality of foster parenting (r = .24, p < .001), whereas somewhat smaller effects were found for non-kinship care, age of the child, placement without siblings and maltreatment of the child. The risks identified in the present meta-analysis were generally modest, but showed generalizability across continent, gender and age.

(4)

4

1. Introduction

Children of parents who are no longer able to raise them may be referred to foster care (Chamberlain et al., 2006; Strijker & Zandberg, 2005), which is intended to improve outcomes for children who tend to be at risk for severe psychopathology (Bilaver, Jaudes, Koepke, & George, 1999), and would otherwise experience seriously inadequate parenting. It is a form of care in which a juvenile is placed in a different environment than the family of origin (Rock, Michelson, Thomson, & Day, 2015; Strijker & van Oijen, 2008). Foster care can be seen as an alternative to living at home or in residential care (Shore, Sim, Le Prohn, & Keller, 2002). Research shows that children enter foster care for several reasons. The largest percentage is placed because of inadequate parenting, such as child abuse and neglect, which may have resulted in severe mental health problems (Oswald, Heil, & Goldbeck, 2010), while much smaller percentages of children are placed because of parent psychopathology, parent delinquency, addiction problems, divorce and death of one parent (McDonald & Brook, 2009; Okma-Rayzner, 2006; Takayama, Wolfe, & Coulter, 1998).

There are different types of foster care, including emergency foster care, respite foster care, treatment foster care, short-term foster care and long-term foster care, which may last several months of years, for children whose home situation does not provide sufficient educational resources during shorter or longer periods of time (Leathers, 2006; Smith, Stormshak, Chamberlain, & Whaley, 2001). Notably, children placed in long-term foster care experience a premature end of their stay in 20 to 50 percent of all cases (Farmer, Moyers, & Lipscombe, 2005; Leathers, 2006; Minty, 1999), having a negative effect on their developmental outcomes (Aarons et al., 2010; Herrenkohl, Herrenkohl, & Egolf, 2003) and well-being (Rubin, O’Reilly, Luan, & Localio, 2007). Our review study focuses on long-term foster care, providing a quantitative overview of factors that may decrease permanency of foster care placements, that is, placement instability.

(5)

5 A recent meta-analysis showed that, on average, children in foster care did not improve in their adaptive functioning and psychopathology (Goemans, van Geel, & Vedder, 2015), which is thought to be related to quality of the foster care environment (Maaskant, van Rooij, & Hermans, 2014), in particular with respect to placement instability (e.g., Aarons et al., 2010). Bronfenbrenner assumed that stability and complexity of the environment are necessary preconditions for education and a positive development of children (Schulze, 2000). Apart from stability and complexity of the child rearing environment, attachment research has shown it is necessary that those who care for a child should do so in a way that reflects the child’s needs in a responsive and sensitive way (Bowlby, 1988). The opposite of stability is instability. Instability has a negative impact on child development in general. Stability not only is a necessary precondition for positive child development, but it is also crucial for the success of a foster care placement (Newton, Litrownik, & Landsverk, 2000).

Instability can be operationalized by placement breakdown or frequency of placement moves (Rock et al., 2015). The term disruption is often preferred by social workers to the term breakdown when placements end abruptly, the reason being that the (unplanned) ending of a placement may not be an unmitigated disaster for a child, but a necessary step towards a more positive living arrangement (Minty, 1999). Breakdown of foster care has been defined as the situation in which one of the involved parties terminates the intervention before having achieved the goals established in the case plan (López López, del Valle, Montserrat, & Bravo, 2011). Disruption or breakdown constitute a special case of foster care placement instability.

The qualitative (systematic) review of Rock et al. (2015) examined correlates of placement moves and breakdown. Correlates of placement instability with the strongest evidence included older age children, externalizing behavior, longer time in care, residential care as first placement setting, separation from siblings, non-kinship foster care compared to kinship foster care, and having multiple social workers. Key protective factors included

(6)

6 placements with siblings, placements with older foster carers, more experienced foster-carers with strong parenting skills and placements where foster-carers provide opportunities for children to develop intellectually (Holland, Faulkner, & Perez-del-Aguila, 2005; Munro & Hardy, 2007).

The quantitative (selective) review of Oosterman, Schuengel, Slot, Bullens and Doreleijers (2007) examined risk and protective factors associated with disruption of foster care placements, that is, placement instability. Some of their findings are consistent with results from Rock et al. (2015), such as placement of older age children, residential care and behavior problems, externalizing was most consistent. Behavior problems appeared a reasonably robust predictor of placement breakdown, especially when other factors were controlled for. Children with a background of residential care had more placement breakdowns. Unexpectedly, Oosterman et al. (2007) did not find non-kinship care (compared to kinship care) to be associated with higher rates of placement disruption (i.e., instability). In contrast to Rock et al. (2015) who found that non-kinship shows more instability in foster care. Oosterman et al. (2007) showed that foster parents who provide high quality caregiving, who are effective at setting boundaries, who are tolerant, emotionally involved and child-centered constitute a protective factor for placement disruption, which is largely in line with the review by Rock et al. (2015). Other protective factors were motivation of the foster parents, sufficient family resources, and support from relatives or caseworkers.

The aim of this study is to conduct a multilevel meta-analysis on (risk) factors associated with placement instability in foster care over the past 25 years, building on the systematic review of Rock et al. (2015) and the meta-analysis by Oosterman et al. (2007). Use is made of new meta-analytic tools, which provide maximum statistical power, the possibility to test more (moderating) factors associated with placement instability than can be achieved in a standard meta-analysis, and thereby prevent loss of information.

(7)

7 Factors are classified in child and caregiver related factors associated with placement instability: (1) Caregiver-related factors are quality of foster parenting, child’s prior out-of-home care episodes, child’s previous number of out-of-out-of-home placements and non-kinship care; (2) Child related factors are age, gender, ethnicity, maltreatment of the child, behavioral problems of the child and placement without siblings (Cross, Koh, Rolock, & Eblen-Manning, 2013). Study characteristics are the moderators: the percentage of females, average age of the child, the continent in which a primary study was performed, and percentage of ethnic minority in the sample. These moderators are included in order to examine whether generalization is warranted to ethnicity, continent, sex and age. Different continents as Europe, USA/Canada and Australia are included because risk factors by continent may have a different effect on instability of foster care placements. Studies examining predictors of foster care instability have used different research designs. Most important is whether studies were prospective or retrospective.

2. Method

2.1 Sample of studies

We selected studies of the meta-analysis of Rock et al. (2015) and Oosterman et al. (2007), including only studies from 1990 to 2016. Studies had to describe original data, and should include factors that may be associated with instability of foster care placement, including breakdown (foster care placement is ended before goals have been achieved) and disruption (foster care placements end abruptly and unplanned). Only articles in peer-reviewed scientific journals, written in English, were searched in electronic databases: PsycINFO and Google Scholar. The following two search strings were combined: (foster care OR out-of-home-care OR out-of-home placement) AND (breakdown OR failure OR disruption OR (in)stability OR continuity OR permanency OR movement OR transition). To determine whether the retrieved studies could be included in our meta-analysis, we read titles,

(8)

8 abstracts and full article texts. In total our literature search strategy yielded 359 studies. After thoroughly screening these studies, we finally found 38 studies that met our inclusion criteria.

2.2 Publication bias

When conducting a meta-analysis, there is a risk of publication bias. Articles reporting non-significant results are less likely to be published than articles reporting significant results. To examine the possibility of publication bias, we conducted funnel plot analysis as described by Duval and Tweedie (2000a, 2000b) by using the function “trimfill” of the metafor package (Viechtbauer, 2010) in the R environment (Version 3.2.0; R Core Team, 2015). If there is no publication bias, the distribution of effect sizes is shaped as a symmetrical funnel, with the standard error on the y-axis and r (the observed effect size) on the x-axis. Among the available techniques for assessing the possibility of publication bias in a meta-analysis, the trim and fill method provides an estimate of the degree to which publication bias might affect the overall mean effect size (Nakagawa & Santos, 2012). In short, the trim and fill method restores the symmetry of an asymmetric funnel plot by imputing missing effect sizes that are calculated on the basis of existing effectsizes.

2.3 Coding of studies

In developing a coding form, guidelines proposed by Lipsey and Wilson (2001) were followed. The variable of most interest in the present meta-analysis is the domain in which a predictor for placement instability can be classified. Coding this variable was needed for examining the effects of different risk domains for placement instability. In order to classify risk factors into domains, we made a classification that was based on putative risk factors for instability, breakdown and disruption.

The following 10 risk domains, based on at least 5 studies assessing these risks, were distinguished in our classification of predictors of instability: Age of the child; Gender (males versus females); Ethnic minority of the child (Ethnic minority versus ethnic majority or

(9)

9 Caucasian white); Maltreatment of the child (a child’s history of neglect, emotional abuse, sexual abuse, or physical abuse); Behavioral problems of the child (internalizing problems, such as anxiety and depression, and externalizing problems, such as aggression and delinquent behavior); Child’s prior out-of-home care episodes; Child’s previous number of out-of-home placements; Placement without siblings; Low quality of foster parenting; Non-kinship care (versus Non-kinship care).

Besides classifying risks for placement instability, we coded several study characteristics that may moderate the association between putative risk factors and placement instability, including gender and age of the child, continent in which the study was performed, ethnicity of the sample, type of behavioral problems, and operationalization of instability (see Table 3). Instability (unqualified) is the number of foster care placements, breakdown is ending placement before having achieved the goals, and disruption is the unanticipated (sudden and unplanned) ending of a foster care placement (López López, et al., 2011; Minty, 1999; Rock et al., 2015).

2.4 Statistical analyses

We calculated the correlation coefficient (r) because we were interested in the association between putative risks for placement instability and factual instability. All statistics were converted to the correlation coefficient (r), and subsequently transformed in Fischer z-scores to be analyzed. Since extreme effect sizes may have a disproportionate influence on conclusions drawn from statistical analyses, we checked for outliers by searching for effect sizes with standardized scores larger than 3.29 or smaller than −3.29 (Tabachnik & Fidell, 2013). Two effect sizes in two different domains of risk factors (behavioral problems of the child and ethnicity) were identified with a z value exceeding 3.29. These two effect sizes were considered outliers. To reduce the impact of these outliers, the raw r values of the outliers were substituted by a new r value that equaled the highest (or lowest) effect that fell

(10)

10 within the normal range in the distribution of effect sizes. In this way a disproportionate influence of the outlying cases on the results of the statistical analyses was reduced.

Most studies reported on multiple risk factors, and therefore more than one effect size could be extracted from these studies. Therefore, a multilevel random effects model, including three levels, was used for the calculation of combined effect sizes and for the moderator analyses in order to account for statistical dependency of effect sizes (Hox, 2002; Van den Noortgate & Onghena, 2003). While level 1 is just random sampling error, level 2 accounts for variance within studies, and level 3 for variance between studies (Assink et al., 2015). This model was used to obtain an overall estimate of the effect of each domain and in case of significant variation between effect sizes from the same study and/or between studies, it was subsequently extended by including moderators to determine whether this variation can be explained by within or between study characteristics. Since each domain comprised qualitatively different predictors for instability of foster care, we conducted separate meta-analyses for the 10 risk domains.

For the statistical analyses we used the function “rma.mv” of the metafor package (Viechtbauer, 2010) in the R environment (version 3.2.0; R Core Team, 2015). The R syntax we used was written so that the three sources of variance as described by for instance Van den Noortgate, López-López, Marin-Martinez, and Sánchez-Meca (2013, 2014) were modeled (Wibbelink & Assink, 2015). The t-distribution was used for testing individual regression coefficients of the meta-analytic models and for calculating the corresponding confidence intervals (Knapp & Hartung, 2003). When models were extended with categorical moderators consisting of three or more categories, the omnibus test of the null hypothesis that all group mean effect sizes are equal, followed an F-distribution. To determine whether the variance between effect sizes from the same study (Level 2), and the variance between studies (Level 3) were significant, the deviance of the full model was compared to the deviance of a model

(11)

11 excluding one of the variance parameters. The sampling variance of observed effect sizes (Level 1) was estimated by using the formula of Cheung (2014). All model parameters were estimated using the restricted maximum likeihood estimation method and before moderator analyses were conducted, each continuous variable was centered around its mean and dichotomous dummy variables were created for all categorical variables. The log-likelihood-ratio-tests were performed one-tailed and all other tests were performed two-tailed. We considered p-values greater than.05 as statistically significant.

3. Results

3.1. Descriptives, central tendency and variability, and assessment of missing data

The present study describes 38 studies (k) from 1990 to 2016. The total sample of N = 55415 foster care children and the size of the samples described in the included studies (at start of the study) ranged from 19 to 16170 participants. The mean age of the participants at start of the study was 8.95 years (SD = 2.33). Studies were conducted in Northern America, including the USA and Canada (k = 21), Europe (k = 16), and in Australia (k = 1). In total, the coded studies produced 182 separate effect sizes, each reflecting the effect of a risk factor for instability in foster care placement.

An overview of the overall effect sizes of the 10 risk factor domains is presented in Table 1. Each overall effect size represents the effect of a risk domain on instability in foster care placement. Significant overall effect sizes were found for six domains, with small (r = 0.150 for maltreatment of the child) to medium-to-large (r = 0.333 for behavioral of the child) effect sizes, based on criteria for the interpretation of the magnitude of effect sizes as formulated by Mullen (1989). No significant overall effect sizes were found for gender, ethnicity, child’s previous number of out-of-home placements and child’s prior out-of-home care episodes, meaning that these effect sizes did not significantly deviate from zero.

(12)

12 The results of the likelihood-ratio tests showed that there was significant variance between effect sizes from the same study (i.e., level 2 variance) in seven risk domains and that there was significant variance between studies (i.e., level 3 variance) in two risk domains (see Table 1). We therefore conducted moderator analyses on 7 risk domains in order to find within or between study characteristics that can explain level 2 or level 3 variance. There was no significant level 2 or level 3 variance in gender, placement without siblings and child’s prior out-of-home care episodes. We therefore did not perform moderator analyses for these risk domains.

The trim and fill analyses indicated that bias was present in 6 of the 10 risk factor domains, as derived from an asymmetrical distribution of effect sizes. Therefore “corrected” overall effect sizes were calculated for these risk domains (see Table 2). After trim and fill analyses, the overall effect sizes for ten domains were significant and ranged from a small effect (r = 0.090 for behavioral problems of the child) to a medium effect (r = 0.241 for low quality of foster parenting) according to the criteria of Mullen (1989). The overall effect sizes for gender, child’s previous number of out-of-home placements and child’s prior out-of-home care episodes were still not significant, and the overall effect sizes for low quality of foster parenting, non-kinship care, age, placement without siblings and maltreatment of the child were reduced to non-significant. For each risk domain, the funnel plot of effect sizes against the standard error is presented in Appendix E.

3.2. Moderator analyses

The results of all moderator analyses are presented in Table 3, where moderators are classified into sample descriptors, research design descriptors, and risk factor characteristics. Below, the moderators tested are described by domain of risk factors in which effect sizes proved to be heterogeneous (i.e., significant level 2 and/or level 3 variance; see Table 1). It

(13)

13 should be noted that level 2 and 3 variances of gender, placement without siblings and child’s prior out-of-home care episodes were homogeneous.

3.2.1. Behavioral problems of the child

We only found a moderating effect for the subdomain of behavioral problems. Externalizing problems yielded a larger effect than internalizing problems and overall problems. No significant moderating effect was found for percentage of females, percentage of minorities, overall mean age, type of continent and type of instability.

3.2.2. Low quality of foster parenting

For low quality of foster parenting, we only found that effects of risk factors for instability in foster care placement were smaller when the percentage of minorities increased. No significant moderating effect was found for percentage of females, overall mean age, type of continent and type of instability.

3.2.3. Non-kinship care

None of the variables tested (i.e., percentage of females, percentage of minorities, overall mean age, type of continent and type of instability) significantly moderated the effect of non-kinship care.

3.2.4. Older age

A moderating effect was found for type of instability. Breakdown yielded a larger effect than instability and disruption. No significant moderating effect was found for percentage of females, percentage of minorities, overall mean age and type of continent.

(14)

14

3.2.5. Maltreatment of the child

None of the variables tested (i.e., percentage of females, percentage of minorities, overall mean age, type of continent and type of instability) significantly moderated the effect of maltreatment of the child.

3.2.6. Ethnic minorities

We only found a moderating effect for type of instability. Disruption yielded a larger effect than instability and breakdown. No significant moderating effect was found for percentage of females, percentage of minorities, overall mean age and type of continent.

3.2.7. Child’s previous number of out-of-home placements

None of the variables tested (i.e., percentage of females, percentage of minorities, overall mean age, type of continent and type of instability) significantly moderated the effect of child’s previous number of out-of-home placements.

(15)

15

Table 1

Results for the Overall Mean Effect Sizes of the 10 Risk Domains

Domain of risk factors # Studies # ES Mean r (SE) 95% CI Sig. Mean r (p) % Var. at level 1 Level 2 variance % Var. at level 2 Level 3 variance % Var. at level 3 (1) Gender (2) Behavioral problems of the child

(3) Low quality of foster parenting

(4) Non-kinship care (5) Older age

(6) Placement without siblings

(7) Maltreatment of the child (8) Ethnic minority (child) (9) Child’s previous number

of out–of-home placements

(10) Child’s prior out-of-home care episodes

12 22 8 13 16 8 5 10 6 5 13 41 12 21 27 10 18 23 11 6 0.427 (0.043) 0.333 (0.049) 0.241 (0.047) 0.206 (0.086) 0.196 (0.060) 0.189 (0.049) 0.150 (0.047) 0.101 (0.058) 0.063 (0.137) -0.020 (0.047) -0.067, 0.118 0.234, 0.431 0.138, 0.345 0.027, 0.385 0.071, 0.320 0.077, 0.300 0.050, 0.250 -0.019, 0.220 -0.242, 0.369 -0.142, 0.102 .563 <.001*** <.001*** .026* .003** .004** .006** .095 .654 .691 3.1 3.4 4.7 0.8 0.5 14.1 11.4 0.5 1.9 12.8 .000 .064*** .001*** .007*** .027*** .002 .029*** .058*** .028*** .003 2.9 83.1 9.8 7.1 42.8 8.9 86.2 90.2 23.2 31.4 .019 .010 .012 .087*** .036 .014 .001 .006 .090* .006 96.9 13.5 85.5 92.1 56.7 77.0 2.4 9.3 74.8 55.8 Note. #studies = number of studies; # ES = number of effect sizes; SE = standard error; CI = confidence interval; Sig = significance; Mean r = mean effect size (r);

% Var = percentage of variance explained; Level 2 variance = variance between effect sizes from the same study; Level 3 variance= variance between studies. * p <.05.

** p <.01. *** p <.001.

(16)

16

Table 2

Results for the Overall Mean Effect Sizes of the 10 Risk Domains After Conducting Trim and Fill Analyses

Domain of risk factors # Studies # ES Mean r (SE) 95% CI Sig. Mean r (p) (1) Gender

(2) Behavioral problems of the child (3) Low quality of foster parenting (4) Non-kinship care

(5) Older age

(6) Placement without siblings (7) Maltreatment of the child

(8) Ethnic minority status of the child (9) Child’s previous number of out-of-home

placements

(10) Child’s prior out-of-home care episodes

- 27 - 16 - 10 - 13 8 6 - 47 - 25 - 12 - 28 14 7 - 0.090 (0.025) - 0.021 (0.013) - -0.001 (0.003) - 0.164 (0.058) 0.022 (0.013) 0.000 (0.002) - 0.039, 0.141 - -0.005, 0.047 - -0.008, 0.007 - 0.045, 0.284 -0.006, 0.049 -0.004, 0.004 - <.001*** - .108 - .858 - .009** .112 .882 Note. #studies = number of studies; # ES = number of effect sizes; Mean r = mean effect size (r); SE = standard error; CI = confidence interval; Sig = significance. Dashes indicate a symmetrical distribution of effect sizes in a risk domain, meaning that trimming and filling of effect sizes was not necessary.

(17)

17

Table 3

Results for Continuous and Categorical Moderators (Bivariate Models)

Moderator variables # Studies # ES Intercept(95% CI) / Mean r (95% CI)

β (95% CI) F (df1, df2)a pb Level 2 variance

Level 3 variance

(2) Behavioral problems of the child Sample descriptors

Percentage of females Percentage of minorities

Risk factor characteristics

Overall mean age Subdomain of risk factors

Internalizing and/or externalizing problems (RC)

Internalizing Externalizing

Research design descriptors

Encoded continent USA (Canada) (RC) Europe Australia Type of instability Instability (RC) Breakdown Disruption

(3) Low quality of foster parenting Sample descriptors

Percentage of females Percentage of minorities

Risk factor characteristics

Overall mean age

Research design descriptors

Encoded continent USA (Canada) (RC) Europe Type of instability Instability (RC) Breakdown Disruption 17 14 13 17 4 7 13 8 1 5 6 11 5 4 4 3 5 3 3 2 35 25 29 23 4 14 28 12 1 15 8 18 7 7 5 6 6 5 5 2 0.349 (0.232, 0.467)*** 0.307 (0.182, 0.433)*** 0.357 (0.212, 0.502)*** 0.235 (0.138, 0.333)*** 0.231 (-0.001, 0.463) 0.600 (0.461, 0.739)*** 0.369 (0.248, 0.491)*** 0.255 (0.081, 0.430)** 0.408 (-0.163, 0.979) 0.351 (0.163, 0.539)*** 0.380 (0.168, 0.592)*** 0.295 (0.146, 0.444)*** 0.221 (0.029, 0.414)* 0.178 (0.107, 0.248)** 0.219 (-0.148, 0.586) 0.216 (0.033, 0.399)* 0.257 (0.113, 0.401)** 0.226 (0.030, 0.423)* 0.290 (0.090, 0.490)** 0.192 (-0.051, 0.435) 0.006 (-0.018, 0.031) 0.001 (-0.004, 0.007) -0.019 (-0.082, 0.044) -0.004 (-0.256, 0.247) 0.364 (0.195, 0.534)*** -0.114 (-0.327, 0,099) 0.038 (0.545, 0.622) 0.029 (-0.254, 0.313) -0.056 (-0.296, 0.184) -0.014 (-0.061, 0.033) 0.005 (0.000, 0.010)* -0.000 (-0.091, 0.091) 0.041 (-0.191, 0.274) 0.064 (-0.217, 0.344) -0.034 (-0.347, 0.278) F(1, 33) = 0.260 F(1, 23) = 0.172 F(1, 27) = 0.382 F(2, 38) = 9.957 F(1, 38) = 0.642 F(2, 38) = 0.253 F(1, 5) = 0.613 F(1, 5) = 7.512 F(1, 3) = 0.000 F(1, 10) = 0.158 F(2, 9) = 0.272 .613 .682 .542 <.001*** .541 .778 .469 .041* .991 .700 .768 .070*** .067*** .084*** .049*** .067*** .065*** .001*** .001*** .000 .001*** .001*** .014 .010 .016 .000 .009 .013 .022 .000 .045 .015 .017

(18)

18 (4) Non-kinship care

Sample descriptors

Percentage of females Percentage of minorities

Risk factor characteristics

Overall mean age

Research design descriptors

Encoded continent USA (Canada) (RC) Europe Type of instability Instability (RC) Breakdown Disruption (5) Older age Sample descriptors Percentage of females Percentage of minorities

Risk factor characteristics

Overall mean age

Research design descriptors

Encoded continent USA (Canada) (RC) Europe Australia Type of instability Instability (RC) Breakdown Disruption

(7) Maltreatment of the child Sample descriptors

Percentage of females Percentage of minorities

Risk factor characteristics

Overall mean age

Research design descriptors

8 6 6 7 6 6 3 4 13 11 7 9 6 1 5 6 5 4 3 4 9 11 7 14 7 13 4 4 22 20 9 16 10 1 9 12 6 12 6 12 0.229 (-0.102, 0.561) 0.203 (-0.275, 0.682) 0.333 (0.076, 0.591)* 0.221 (-0.032, 0.474) 0.189 (-0.090, 0.467) 0.219 (-0.053, 0.490) 0.347 (-0.045, 0.739) 0.080 (-0.262, 0.421) 0.199 (0.048, 0.350)* 0.140 (0.031, 0.249)* 0.210 (-0.102, 0.522) 0.112 (-0.049, 0.273) 0.351 (0.143, 0.559)** 0.078 (-0.448 0.604) 0.219 (0.043, 0.394)* 0.326 (0.163, 0.489)*** -0.003 (-0.196, 0.190) 0.199 (0.037, 0.361)* 0.131 (-0.122, 0.385) 0.207 (0.050, 0.364)* 0.029 (-0.047, 0.104) 0.002 (-0.053, 0.056) -0.096 (-0.201, 0.010) -0.032 (-0.409, 0.344) 0.128 (-0.349, 0.605) -0.139 (-0.575, 0.297) 0.016 (-0.014 0.046) 0.000 (-0.005, 0.006) 0.099 (-0.119, 0.317) 0.239 (-0.024, 0.502) -0.034 (-0.584, 0.516) 0.108 (-0.132, 0.347) -0.222 (-0.482, 0.039) -0.011 (-0.051 0.029) 0.002 (-0.030 0.033) 0.041 (-0.045, 0.128) F(1, 8) = 0.775 F(1, 9) = 0.005 F(1, 5) = 5.426 F(1, 19) = 0.033 F(2, 18) = 0.591 F(1, 20) = 1.278 F(1, 18) = 0.015 F(1, 7) = 1.147 F(2, 24) = 1.864 F(2, 24) = 3.660 F(1, 6) = 2.551 F(1, 4) = 1.557 F(1, 3) = 0.004 .404 .943 .067 .859 .564 .272 .903 .320 .177 .041* .161 .280 .956 .008** .006*** .000 .007*** .007*** .031*** .028*** .000 .024*** .029*** .001 .002 .006 .150** .239*** .052 .096*** .095*** .041 .010 .134* .037 .016 .000 .002 .006

(19)

19 Encoded continent USA (Canada) (RC) Europe Type of instability Instability (RC) Breakdown Disruption (8) Ethnic minority Sample descriptors Percentage of females Percentage of minorities

Risk factor characteristics

Overall mean age

Research design descriptors

Encoded continent USA (Canada) (RC) Europe Type of instability Instability (RC) Breakdown Disruption

(9) Child’s previous number of out-of-home placements

Sample descriptors

Percentage of females Percentage of minorities

Risk factor characteristics

Overall mean age

Research design descriptors

Encoded continent USA (Canada) (RC) Europe Type of instability Instability (RC) Breakdown Disruption 2 3 2 1 2 8 9 4 8 2 3 1 6 4 3 2 5 1 1 1 4 10 8 7 6 5 16 19 8 21 2 10 1 12 8 7 3 10 1 4 2 5 0.196 (0.058, 0.333)** 0.103 (-0.035, 0.240) 0.255 (0.077, 0.433)** 0.085 (-0.072, 0.242) 0.127 (-0.042, 0.297) 0.105 (-0.044, 0.254) 0.120 (-0.029, 0.270) 0.221 (-0.103, 0.544) 0.099 (-0.031, 0.230) 0.119 (-0.270, 0.507) -0.044 (-0.194, 0.106) 0.041 (-0.429, 0.511) 0.220 (0.080, 0.361)** 0.008 (-0.246, 0.263) -0.128 (-0.698, 0.441) -0.290 (-2.812, 2.231) 0.059 (-0.312, 0.429) 0.091 (-0.784, 0.966) 0.177 (-0.453, 0.808) 0.489 (-0.188, 1.167) -0.091 (-0.448, 0.265) -0.093 (-0.287, 0.102) -0.170 (-0.407, 0.068) -0.127 (-0.374, 0.119) 0.022 (-0.007, 0.050) -0.001 (-0.011, 0.008) 0.001 (-0.181, 0.182) 0.019 (-0.391, 0.429) 0.085 (-0.408, 0.578) 0.264 (0.059, 0.470)* -0.045 (-0.091, 0.001) -0.012 (-0.049, 0.025) 0.160 (-1.264, 1.584) 0.032 (-0.918, 0.982) 0.312 (-0.614, 1.238) -0.269 (-0.993, 0.456) F(1, 8) = 0.119 F(1, 8) = 1.326 F(1, 14) = 2.644 F(1, 17) = 0.110 F(1, 6) = 0.000 F(1, 21) = 0.009 F(2, 20) = 3.621 F(1, 6) = 5.694 F(1, 5) = 0.739 F(1, 1) = 2.035 F(1, 9) = 0.006 F(2, 8) =1.639 .739 .283 .126 .744 .994 .923 .046* .054 .429 .389 .941 .253 .001 .002 .074*** .073*** .136*** .060*** .051*** .031*** .033*** .096*** .029*** .028*** .016 .011 .000 .008 .000 .007 .000 .020 .123 .012 .113* .067

(20)

20

Note.# studies= number of studies; # ES = number of effect sizes; mean r = mean effect size (r); CI = confidence interval; β = estimated regression coefficient; Level 2 variance = variance between effect sizes from the same study; Level 3 variance= variance between studies.

a Omnibus test of all regression coefficients in the model.

b p-Value of the omnibus test.

* p <.05.

** p <.05. *** p <.001

(21)

21

4. Discussion

This review aimed to generate more specific knowledge on the association between different types of risk factors and foster placement instability by summarizing the effects of risk factors in a meta-analysis. More specifically, we quantitatively examined the effects of several risk domains for instability of foster care placements, including disruption and breakdown. Furthermore, we examined whether the effects of these risk domains were moderated by sample characteristics, research design and risk factors.

4.1 Overall and moderator effects of domains of risk factors

A significant overall effect size was found for 6 of the 10 domains of risk factors, which ranged from r = 0.150 (for maltreatment of the child) to r = 0.333 (for behavioral problems of the child). We found that behavioral problems of the child, low quality of foster parenting, non-kinship care, age of the foster child, placement without siblings and maltreatment of the child were significantly associated with instability of foster care placements. Gender, ethnicity, child’s previous number of out-of-home placements and child’s prior out-of-home care episodes showed no significant overall effect sizes, meaning that these risk domains were not associated with instability of foster care placements. In general, these findings confirm that exposure to different types of risk factors increases the chance for instability of foster care placements, indicating that multiple risk domains are involved in placement instability (Oosterman et al., 2007; Rock et al., 2015).

The strongest effect was found for behavioral problems of the child, in particular externalizing problems (r = 0.600), yielding a very large effect. This result may not be surprising, as externalizing problems cause parenting stress and are difficult to handle for parents (Holland & Gorey, 2004; Wilson, 2006). Externalizing problem behavior of children has been shown to have a negative direct impact on supportive parenting behaviors and positive involvement of the foster parent (Holland & Gorey, 2004). Foster parents

(22)

22 experienced more parenting burden and parenting stress, which can lead to instability of foster care placements. Moreover, there is empirical evidence showing that (especially externalizing) behavioral problems remain stable or can even increase during foster care placement; a decrease is rarely observed (Bastiaensen, 2001; Van Holen, Vanderfaeillie & Trogh, 2007; Wilson, 2006).

The next strongest effect was found for low quality of foster parenting, moderated by the percentage of ethnic minorities in the sample, indicating that low quality of foster parenting was stronger related to placement instability in children from ethnic minority background, which in itself constitutes a risk factor. This may be explained by the multiple-risk model, which assumes that problems mainly arise when multiple multiple-risk factors are present simultaneo1usly (Elder & Caspi, 1988; Hermann, 2007; Rutter, 2008; Sampson & Laub, 1994). The underlying theory is that an excess of stressors leads to a disruption of the self-regulation of family members (especially parents) and the family as a whole (Evans, 2003; Evans & English, 2002; Hermanns, 1987; Hermanns 1998). Especially families living in a socially disadvantaged situation are more likely to face multiple risks. Ethnic minorities have more often to deal with social deprivation, living in poor neighborhoods, discrimination and low SES. Accumulation of these risks may lead to more instability in ethnic minorities (Garland et al., 2000). It is plausible to suggest that multi-model family-based interventions, which target inadequate parenting and risks associated with ethnic minority status, may prevent placement instability in foster families with an ethnic minority background.

Ethnicity did only have an effect on disruption, that is, when foster care placements end abruptly and unplanned (Minty, 1999) instead of general instability or placement breakdown, which occurs when goals have not yet been achieved (López López et al., 2011). It is difficult to explain why the effect of ethnicity status did only affect disruption. A first explanation would be that foster parents with an ethnic minority background have a relatively

(23)

23 high threshold for care utilization. A second explanation might be that children from ethnic minorities with the most serious problems may in particular be placed in non-kinship foster care families, where lack of goodness of fit due to cultural differences may increase the risk for disruption (Bakker, 2014; Winokur, Holtan, & Batchelder, 2014). It seems therefore desirable to account for the cultural background of the child in foster care placements (Brown, George, Sintzel, & Arnault 2009; Williams & Soydan 2005). Especially in the United States, much research has been conducted on the benefits of cultural matching in foster care (Sinclair, Wilson & Gibbs, 2003). Becker, Jordan and Larsen (2007) showed that if there are no shared values, conflicts may arise between the foster child and his or her foster parents. The risk of adjustment problems is greater when foster children are not able to maintain their cultural identity and family ties (Strijker, Zandberg, & Van der Meulen, 2001). This could be a cause for disruption in foster care placements. In this context, the distinction between kinship and non-kinship placements seems relevant. Research shows that kinship foster care may have some advantages compared to non-kinship foster care, such as continuity in living conditions (if safety is guaranteed), a similar cultural background, and lower risk for stigmatization (Holtan, 2005).

Placement without siblings and maltreatment of the child have an effect on placement instability. Placement without siblings is associated with more placement breakdown. Barth et al. 2007 reported that children feel less secure when separated from their siblings and report missing them as much as their parents which increases the risk of placement instability. Children were more likely to experience several placements when the reason for placement was the child's behavior problems. Behavior related placement changes are associated with emotional abuse, rather than neglect or sexual or physical abuse(James, 2004). This association increases the risk of placement instability.

(24)

24 Oosterman et al. (2007) did not find non-kinship care (compared to kinship care) to be associated with higher rates of placement disruption (i.e., instability). In contrast to Rock et al. (2015), who found that non-kinship showed more instability in foster care. We found a medium effect (r = 0.206) for non-kinship care, which means that non-kinship care is associated with higher rates of placement instability. A first explanation would be that children in non-kinship care show more psychopathology than children in kinship foster care (this study; Oosterman et al., 2007; Rock et al., 2015; Winokur et al., 2014). A second explanation is that non-kinship foster parents have been shown to be less dedicated and personally involved than kinship foster parents (Honomichl & Brooks, 2010; Le Prohn, 1994).

Age did have an effect on placement instability, including breakdown, but not disruption. Older children often have more behavioral problems (Barth et al., 2007). Older children need less education and are just looking for autonomy, which can cause more breakdown (Berridge, 1997; Pardeck, 1984; Rowe, Cain, Hundleby, & Garnett, 1989). It is plausible to suggest that an adolescent’s educational difficulties plus the onset of the typical changes of this stage might make it more likely for the families to decide to put an end to placement at these ages, even when educational goals have not yet been achieved (Berridge, 1997; Napier, 1972; Pardeck, 1984; Rowe et al., 1989).

We found no significant effect of child’s prior out-of-home care episodes and child’s previous number of out-of-home placements, which means that length of previous care and a history of instability are static factors that do not predict future instability of foster care placements. In fact, the present study shows that dynamic risk factors have more explanatory power than static risk factors, which is a promising result from the perspective of successful prevention of out-of-home placements, because only dynamic risk factors can be targeted in preventive interventions.

(25)

25 Oosterman et al. (2007) concluded that a history of residential care was a predictor of breakdown. However, this result was based on only one study of Cooper, Peterson and Meier (1987) examining a group of abused and/or neglected children placed in a residential home for maltreated children. Also, Rock et al. (2015) showed a history of residential care to be a risk is for placement instability, but this finding was based on only three studies (Barth et al., 2007; Park & Ryan, 2009; Trasler, 1960). In the present meta-analysis, history of residential care was not included as a predictor, because there were not enough studies to examine this moderator in a meaningful way. We examined only moderators based on five or more studies. Finally, Oosterman et al. 2007 found that children with drug or alcohol addicted parents experienced more disruptions in foster care than children without a family history of drug and alcohol abuse, but again this result was based on few studies.

A number of putative moderators did not have an effect on the association between risk factors and placement instability, including continent, overall mean age of the parents and percentage of females. This means that the results may be generalized across continent, ethnicity, gender and age.

4.2 Limitations

Several limitations of the present study should be mentioned. First, the effect size for age may be an underestimation of the true effect, because there are also some indications for a curvilinear effect of age, with the highest risk for the younger children and late adolescents, and a lower risk for school-aged children and early adolescents (Helton, 2011; Joseph, O’Connor, Briskman, Maughan, & Scott, 2014).

Second, most included studies did not examine who wanted to end foster care; the child, foster parents, care providers, or youth protection services. Sallnäs, Vinnerljung and Kyhle Westermark (2004) examined who ended foster care. Between one-third and one-half

(26)

26 of new Swedish placements did end in breakdown, most often initiated by the child or by the care providers.

Third, we found indications for publication bias in 6 of the 10 risk domains, meaning that the true overall effect sizes of these domains may be different from the overall effects that were estimated in the present meta-analytic study. For some domains small effect sizes were underrepresented (the adjusted mean r was smaller than the observed mean r), whereas for most other domains large effect sizes were underrepresented (the adjusted mean r was larger than the observed mean r). However, it is important to note that the adjusted mean effect sizes that were produced by the trim and fill analyses should not be regarded as true mean effects. There are several methodological difficulties regarding this method. For instance, Nakagawa and Santos (2012) argued that trim and fill analysis should meet the assumption of independence of effect sizes, which was violated in our multi-level meta-analytic approach. Terrin, Schmid, Lau, and Olkin (2003) showed that between-study heterogeneity invalidates the results produced by trim and fill analysis (for a similar result, see Peters, Sutton, Jones, Abrams, & Rushton, 2007). Therefore, the differences between the adjusted and the observed mean effect sizes in the present study should be interpreted as indicative of publication or selection bias. This possible bias proved to be considerable for ethnic minority of the child (Δ mean r = 0.063) and child’s prior out-of-home care episodes (Δmean r = 0.020).

We could not examine whether paid or unpaid foster care did have an effect on placement instability, because studies did not report on payment of foster parents. Also, most studies did not examine ethnicity of the foster parents, which limits the examination of how ethnicity of the child can have an impact on placement instability. Notably, it is still unknown which influences cultural or religious matching has for the longer term development of the child and the course of the foster placement (Brown et al., 2009). To further clarify the

(27)

27 relation between ethnicity and instability, it is important to consider ethnicity of the foster parents in future research.

4.3 Conclusion

The present review contributes to the literature on risk factors for instability of foster care placements. The largest effects were found for behavioral problems of the child, although this effect was significantly reduced after correction for publication bias, and low quality of foster parenting, whereas somewhat smaller effects were found for non-kinship care, age of the child, placement without siblings and maltreatment of the child. The risks identified in the present meta-analysis are generally modest, but show generalizability across continent, gender and age. These risks may not only be viable targets in interventions that aim to prevent instability of foster care placements, in particular quality of foster parenting, but may also figure as indicators of future risk for placement instability in actuarial risk assessment instruments that should be developed to validly and reliably predict instability of foster care placements.

(28)

28

References1

Aarons, G. A., James, S., Monn, A. R., Raghavan, R., Wells, R. S., & Leslie, L. K. (2010). Behavior problems and placement change in a national child welfare sample: A prospective study. Journal of the American Academy of Child & Adolescent

Psychiatry, 49(1), 70-80. doi:10.1016/j.jaac.2009.09.005

*Akin, B. A. (2011). Predictors of foster care exits to permanency: A competing risks analysis of reunification, guardianship, and adoption. Children and Youth Services

Review, 33(6), 999-1011. doi:10.1016/j.childyouth.2011.01.008

*Andersen, S. H., & Fallesen, P. (2015). Family matters?. The effect of kinship care on foster care disruption rates. Child abuse & neglect, 48, 68-79. doi:

10.1016/j.chiabu.2015.06.005

Bakker, J. (2014). Differences in psychosocial functioning of foster children between kinship and non-kinship foster care: A meta-analysis. Master Thesis, University of

Amsterdam.

*Barber, J. G., Delfabbro, P. H., & Cooper, L. (2001). The predictors of unsuccessful transition to foster care. Journal of Child Psychology and Psychiatry, 42(6), 785-790. doi:10.1017/S002196300100751

Barth, R. P., Lloyd E. C., Green R. L., James S., Leslie, L. K., & Landsverk J. (2007) Predictors of placement moves among children with and without emotional and

(29)

29 behavioral disorders. Journal of Emotional and Behavioral Disorders, 15(1), 46–55. doi:10.1177/10634266070150010501

Becker, M. A., Jordan, N., & Larsen, R. (2007). Predictors of successful permanency planning and length of stay in foster care: The role of race, diagnosis and place of

residence. Children and Youth Services Review, 29(8), 1102-1113. doi:10.1016/j. childyouth.2007.04.009

Bilaver, L. A., Jaudes, P. K., Koepke, D., & George, R. M. (1999). The health of children in foster care. Social Service Review, 73(3), 401–417. doi:10.1086/514430

Bowlby, J. (1988). A secure base: Parent-child attachment and healthy human development. New York: Basic Books.

Brown, J. D., George, N., Sintzel, J., St. Arnault, D. (2009). Benefits of cultural matching in foster care. Children and Youth Services Review, 31(9), 1019-1024. doi:

10.1016/j.childyouth.2009.05.001

*Chamberlain, P., Price, J. M., Reid, J. B., Landsverk, J., Fisher, P. A., & Stoolmiller, M. (2006). Who disrupts from placement in foster and kinship care?. Child abuse &

neglect, 30(4), 409-424. doi:10.1016/j.chiabu.2005.11.004

Cheung, M. W. L. (2014). Modeling dependent effect sizes with three-level meta-analyses: A structural equation modeling approach. Psychological Methods, 19(2), 211–229. doi:10.1037/a0032968

Cooper, C. S., Peterson, N. L., & Meier, J. H. (1987). Variables associated with disrupted placement in a select sample of abused and neglected children. Child Abuse and

Neglect, 11, 75–86. doi:10.1016/0145-2134(87)90035-4

Cross, T. P., Koh, E., Rolock, N., & Eblen-Manning, J. (2013). Why do children experience multiple placement changes in foster care? Content analysis on reasons for

(30)

30 .751300

*Crum, W. (2010). Foster parent parenting characteristics that lead to increased placement stability or disruption. Children and Youth Services Review, 32(2), 185-190. doi: 10.1016/j.childyouth.2009.08.022

*DeGarmo, D. S., Chamberlain, P., Leve, L., D., & Price, P. (2009). Foster parent intervention engagement moderating child behavior problems and placement disruption. Research on Social Work Practice, 4(19), 423-433. doi:10.1177 /1049731508329407

*Drapeau, S., Simard, M., Beaudry, M., & Charbonneau, C. (2000). Siblings in family transitions. Family Relations, 49(1), 77-85. doi:10.1111/j.1741-3729.2000.00077.x Duval, S., & Tweedie, R. (2000a). A nonparametric ‘trim and fill’ method of accounting for

publication bias in meta-analysis. Journal of the American Statistical Association,

95(449), 89–99. doi:10.1080/01621459.2000.10473905

Duval, S., & Tweedie, R. (2000b). Trim and fill: A simple funnel-plot-based method of testing and adjusting for publication bias in meta-analysis. Biometrics, 56(2), 455-463. doi:10.1111/j.0006-341X.2000.00455.x.

Elder G. H., & Caspi, A. (1988). Economic stress in lives: Developmental

perspectives. Journal of Social Issues, 44, 25–45. doi:10.1111/j.15404560.1988 .tb02090.x

Evans, G. (2003). A multi methodological analysis of cumulative risk and allostatic load among rural children. Developmental Psychology, 39 (5), 924-933. doi:10.1037 /00121649.39.5.924

Evans, G., & K. English (2002). The environment of poverty: multiple stressor exposure, psychophysiological stress and socioemotional adjustment. Child Development, 73(4), 1238-1248. doi:10.1111/1467-8624.00469

(31)

31 Farmer, E., Lipscombe, J., & Moyers, S. (2005). Foster care strain and its impact on parenting

and placement outcomes for adolescents. British Journal of Social Work, 35, 237‐253. doi:10.1093/bjsw/bch181

*Fisher, P. A., Stoolmiller, M., Mannering, A. M., Takahashi, A., & Chamberlain, P. (2011). Foster placement disruptions associated with problem behavior: mitigating a threshold effect. Journal of consulting and clinical psychology, 79(4), 481-484. doi:

10.1037/a0024313

Garland, A. F., Hough, R., Landsverk, J. A., McCabe, K. M., Yeh, M., Ganger, W. C., & Reynolds, B. (2000). Racial/Ethnic differences in mental health care utilization among children in foster care. Children’s Services: Social Policy, Research, and Practice, 3, 133-146. doi:10.1207/S15326918CS0303_1

George, V. (1970). Foster care. Theory and Practice. London: Routledge & Kegan Paul. Goemans, A., Van Geel, M., & Vedder, P. (2015). Over three decades of longitudinal research on the development of foster children: A meta-analysis. Child abuse &

neglect, 42, 121-134. doi:10.1016/j.chiabu.2015.02.003

*Helton, J. J. (2011). Children with behavioral, non-behavioral, and multiple disabilities, and the risk of out-of-home placement disruption. Child abuse & neglect, 35(11), 956-964. doi:10.1016/j.chiabu.2011.06.004

Hermanns, J. (1987). Risicofactoren in de ontwikkeling van kinderen: de rol van de

bijsturende omgeving. Kind en Adolescent, 8, 49-59. Retrieved from http://dare.uva.nl /search?field1=dai;value1=068425651;docsPerPage=1;startDoc=183

Hermanns, J. (1998). Family risk and family support: an analysis of concepts. In: J. Hermanns & H. Leu (red.), Family risks and family support (pp. 9-37). Delft: Eburon.

(32)

32 Herrenkohl, E. C., Herrenkohl, R. C., & Egolf, B. P. (2003). The psychosocial consequences

of living environment instability on maltreated children. American Journal of

Orthopsychiatry, 73(4), 367-380. doi:10.1037/0002-9432.73.4.367

*Holland, P., & Gorey, K. M. (2004). Historical, developmental, and behavioral factors associated with foster care challenges. Child and Adolescent Social Work

Journal, 21(2), 117-135. doi:10.1023/B:CASW.0000022727.40123.95

Holland, S., Faulkner, A., & Perez‐del‐Aguila, R. (2005). Promoting stability and continuity of care for looked after children: A survey and critical review. Child & Family Social

Work, 10(1), 29-41. doi:10.1111/j.1365-2206.2004.00339.x

*Holtan, A., Handegård, B. H., Thørnblad, R., & Vis, S. A. (2013). Placement disruption in long-term kinship and nonkinship foster care. Children and Youth Services

Review, 35(7), 1087-1094. doi:10.1016/j.childyouth.2013.04.022

Honomichl, R. D., & Brooks, S. (2010). Predictors and Outcomes of Long Term Foster Care: A Literature Review. Davis Human Services. Northern California Training Academy. *Hurlburt, M. S., Chamberlain, P., DeGarmo, D., Zhang, J., & Price, J. M. (2010). Advancing

prediction of foster placement disruption using Brief Behavioral Screening. Child

abuse & neglect, 34(12), 917-926. doi:10.1016/j.chiabu.2010.07.003

Hox, J. (2002). Multilevel analysis: Techniques and applications. Mahwah, NJ: Lawrence Erlbaum Associates.

*Iglehart, A. P. (1994). Kinship foster care: Placement, service, and outcome issues. Children and Youth Services Review, 16(1), 107-122. doi:10.1016/0190-7409(94)90018-3 *James, S. (2004). Why do foster care placements disrupt? An investigation of reasons for

placement change in foster care. Social service review, 78(4), 601-627. doi:10.1086 /424546

(33)

33 Joseph, M. A., O' Connor, T. G., Briskman, J. A., Maughan, B., & Scott, S. (2014). The

formation of secure new attachments by children who were maltreated: an observational study of adolescents in foster care. Development and

psychopathology, 26(01), 67-80. doi:10.1017/S0954579413000540

*Kalland, M., & Sinkkonen, J. (2001). Finnish children in foster care: Evaluating the breakdown of long-term placements. Child welfare, 80(5), 513-527. doi: 10.6094575/0009-4021

Knapp, G., & Hartung, J. (2003). Improved tests for a random effects meta-regression with a single covariate. Statistics in Medicine, 22(17), 2693–2710. doi:10.1002/sim.1482 *Leathers, S. J. (2005). Separation from siblings: Associations with placement adaptation and

outcomes among adolescents in long-term foster care. Children and Youth Services

Review, 27(7), 793-819. doi:10.1016/j.childyouth.2004.12.015

*Leathers, S. J. (2006). Placement disruption and negative placement outcomes among adolescents in long-term foster care: The role of behavior problems. Child abuse &

neglect, 30(3), 307-324. doi:10.1016/j.chiabu.2005.09.003

Le Prohn, N. S. (1994). The role of the kinship foster parent: A comparison of the role conceptions of relative and non-relative foster parents. Children and Youth Services

Review, 16(1), 65-84. doi:10.1016/0190-7409(94)90016-7

Lipscombe, J., Farmer, E., & Moyers, S. (2003). Parenting fostered adolescents: Skills and strategies. Child & Family Social Work, 8(4), 243-255. doi:10.1046/j.1365-2206 .2003.00294

Lipsey, M. W., & Wilson, D. B. (2001). Practical meta-analysis. Thousands Oaks, CA: Sage. *López López, M., del Valle, J. F., Montserrat, C., & Bravo, M. (2011). Factors Affecting

Foster Care Breakdown in Spain. The Spanish Journal of Psychology, 14(1), 111-122. doi:10.5209/rev_SJOP.2011.v14.n1.9

(34)

34 *Lutman, E., Hunt, J., & Waterhouse, S. (2009). Placement stability for children in kinship

care: A long-term follow-up of children placed in kinship care through care

proceedings. Adoption & Fostering, 33(3), 28-39. doi:10.1177/030857590903300304 Maaskant, A. M., van Rooij, F. B., & Hermanns, J. M. A. (2014). Mental health and

associated risk factors of Dutch school aged foster children placed in long-term foster care. Children and Youth Services Review, 44, 207-216. doi:10.1016/j.childyouth .2014.06.011

*McAuley, C., & Trew, K. (2000). Children's adjustment over time in foster care: cross informant agreement, stability and placement disruption. British Journal of Social

Work, 30(1), 91-107. doi:10.1093/bjsw/30.1.91

McDonald, T., & Brook, J. (2009). Typologies of Children in Foster Care for Reasons Other than Abuse or Neglect. Journal of Public Child Welfare, 3(4), 391-408. doi:10.1080 /15548730903347861

Minty, B. (1999). Annotation: Outcomes in long-term foster family care. Journal of child

Psychology and Psychiatry, 40(7), 991-999. doi:10.1016/j.chiabu.2005.09.003

Mullen, B. (1989). BASIC meta-analysis: Procedures and programs. Hillsdale, NJ: Lawrence Erlbaum Associates.

Munro, E., & Hardy, A. (2007). Placement Stability: A Review of the Literature.

Loughborough: Loughborough University. Department for Education and Skills. Nakagawa, S., & Santos, E. S. A. (2012). Methodological issues and advances in biological

meta-analysis. Evolutionary Ecology, 26(5), 1253–1274. doi:10.1007/s10682-012 9555-5

*Newton, R. R., Litrownik, A. J., & Landsverk, J. A. (2000). Children and youth in foster care: Disentangling the relationship between problem behaviors and number of

(35)

35 placements. Child abuse & neglect, 24(10), 1363-1374. doi:10.1016/S01452134 (00)00189-7

Okma-Rayzner, C.M. (2006). Pleegkinderen in conflictsituaties (proefschrift). Retrieved from http://dspace.library.uu.nl/handle/1874/8788

*O' Neill, M., Risley-Curtiss, C., Ayón, C., & Williams, L. R. (2012). Placement stability in the context of child development. Children and Youth Services Review, 34(7), 1251-1258. doi:10.1016/j.childyouth.2012.02.018

Oosterman, M., Schuengel, C., Slot, N. W., Bullens, R. A., & Doreleijers, T. A. (2007). Disruptions in foster care: A review and meta-analysis. Children and youth services

review, 29(1), 53-76. doi:10.1016/j.childyouth.2006.07.003

Oswald, S. H., Heil, K., & Goldbeck, L. (2010). History of maltreatment and mental health problems in foster children: A review of the literature. Journal of Pediatric

Psychology, 35(5), 462-472. doi:10.1093/jpepsy/jsp114

*Palmer, S. E. (1996). Placement stability and inclusive practice in foster care: An empirical study. Children and Youth Services Review, 18(7), 589-601. doi:10.1016/0190 7409(96)00025-4

Pardeck, J. T. (1984). Multiple placement of children in foster family care: An empirical analysis. Social Work, 29(6), 506-509. doi:10.1093/sw/29.6.506

Park, J. M., & Ryan, J. P. (2009). Placement and permanency outcomes for children in out-of-home care by prior inpatient mental health treatment. Research on Social Work

Practice, 19(1), 42-51. doi:10.1177/1049731508317276

*Perry, G., Daly, M., & Kotler, J. (2012). Placement stability in kinship and non-kin foster care: A Canadian study. Children and Youth Services Review, 34(2), 460-465. doi: 10.1016/j.childyouth.2011.12.001

(36)

36 Peters, J. L., Sutton, A. J., Jones, D. R., Abrams, K. R., & Rushton, L. (2007). Performance of

the trim and fill method in the presence of publication bias and between-study heterogeneity. Statistics in Medicine, 26(25), 4544–4562. doi:10.1002/sim.2889 R Core Team (2015). R: A language and environment for statistical computing. Vienna,

Austria: R Foundation for Statistical Computing (Retrieved from http://www.Rproject. org/).

Rock, S., Michelson, D., Thomson, S., & Day, C. (2015). Understanding foster placement instability for looked after children: A systematic review and narrative synthesis of quantitative and qualitative evidence. British Journal of Social Work, 45, 177-203. doi:10.1093/bjsw/bct084

Rowe, J., Hundleby, M., & Garnett, L. (1989) Child Care Now: A Survey of Placement

Patterns. BAAF, London.

Rubin, D. M., O' Reilly, A. L., Luan, X., & Localio, A. R. (2007). The impact of placement stabtility on behavioral well-being for children in foster care. Pediatrics, 119(2), 336-344. doi:10.1542/peds.2006-1995

Rutter, M. (2008). Understanding and testing risk mechanisms for mental disorders. Journal

of Child Psychology and Psychiatry, 50, 44-52. doi: 10.1111

/j.14697610.2008.01976.x

*Sallnäs, M., Vinnerljung, B., & Kyhle Westermark, P. (2004). Breakdown of teenage placements in Swedish foster and residential care. Child & Family Social Work, 9(2), 141-152. doi:10.1111/j.1365-2206.2004.00309.x

Sampson, R. J, & Laub, J. H. (1994). Urban poverty and the family context of delinquency: A new look at structure and process in a classic study. Child Development, 65, 523–540. doi:10.1111/j.1467-8624.1994.tb00767.

(37)

37 *Santen, E. (2015). Factors associated with placement breakdown initiated by foster parents

empirical findings from Germany. Child & Family Social Work, 20(2), 191-201. doi: 10.1111/cfs.12068

Schulze, H. J. (2000). Stability and Complexity: Perspectives For a Child-Oriented Family

Policy (pp 9-36). Amsterdam, Oxford, Boston: VU University Press.

Shore, N., Sim, K. E., Le Prohn, N. S., & Keller, T. E. (2002). Foster parent and teacher assessments of youth in kinship and non-kinship foster care placements: Are behaviors perceived differently across settings? Children and Youth Services Review, 24(1), 109-134. doi:10.1016/S0190-7409(01)00170-0

*Sinclair, I., Wilson, K., & Gibbs, I. (2001). A life more ordinary: What children want from foster placements. Adoption and Fostering, 25(4), 17-26. doi:10.1177

/030857590102500404

Smith, D. K., Stormshak, E., Chamberlain, P., & Whaley, R. B. (2001). Placement disruption in treatment foster care. Journal of emotional and behavioral disorders, 9(3), 200-205. doi:10.1177/106342660100900306

*Strijker, J., & Zandberg, T. (2005). Breakdown in foster care. International Journal of Child

and Family Welfare, 8, 76-87.

Strijker, J., Zandberg, T., & Van der Meulen, B. F. (2001). Typologies and outcomes for foster children. Child and Youth Care Forum, 34(1), 43-55. doi:10.1007

/s105660040881-9

*Strijker, J., & Van Oijen, S. (2008). Agreement in foster care. Discrepancies between foster child and foster parent about the severity of problem behavior. International

Journal of Child & Family Welfare, 11, 66‐76.

*Strijker, J., & Van Der Loo, S. (2010). Placement Breakdown Among Foster Children with Intellectual Disabilitities. The British Journal of Developmental Disabilities, 111(56),

(38)

38 111-121.

Tabachnik, B. G., & Fidell, L. S. (2013). Using multivariate statistics (6th ed.). Boston: Allyn and Bacon.

Takayama, J. I., Wolfe, E., & Coulter, K. P. (1998). Relationship between reason for

placement and medical findings among children in foster care. Pediatrics, 101(2), 201 -207. doi:10.1542/peds.101.2.201

Terrin, N., Schmid, C. H., Lau, J., & Olkin, I. (2003). Adjusting for publication bias in the presence of heterogeneity. Statistics in Medicine, 22, 2113–2126 doi:10.1002/sim .1461

*Thorpe, M. B., & Swart, G. T. (1992). Risk and protective factors affecting children in foster care: a pilot study of the role of siblings. The Canadian Journal of Psychiatry/La Revue canadienne de psychiatrie, 37(9), 616-622. Retrieved from: psycnet.apa. org/psycinfo/1993-23070-001

Trasler, G. (1960). In place of parents. London: Routledge & Kegan Paul.

Van den Noortgate, W., López-López, J. A., Marin-Martinez, F., & Sánchez-Meca, J. (2013). Three-level meta-analysis of dependent effect sizes. Behavior Research Methods,

45(2), 576–594. doi:10.3758/s13428-012-0261-6

Van den Noortgate, W., López-López, J. A., Marin-Martinez, F., & Sánchez-Meca, J. (2014). Meta-analysis of multiple outcomes: A multilevel approach. Behavior Research

Methods. doi:10.3758/s13428-014-0527-2 (Advance online publication).

Van den Noortgate, W., & Onghena, P. (2003). Multilevel meta-analysis: A comparison with traditional meta-analytical procedures. Educational and Psychological

(39)

39 Van Holen, F., Vanderfaeillie, J., & Trogh, L. (2007). Longitudinaal onderzoek naar

gezinsbelasting en gedrags- en emotionele problemen in pleegzorg. Tijdschrift voor

Orthopedagogiek, 46, 233-242. Retrieved from http://opvang.be/pdf/publicaties

/publicatie_002.pdf

*Van Rooij, F., Maaskant, A., Weijers, I., Weijers, D., & Hermanns, J. (2015). Planned and unplanned terminations of foster care placements in the Netherlands: Relationships with characteristics of foster children and foster placements. Children and youth

services review, 53, 130-136. doi:10.1016/j.childyouth.2015.03.022

Viechtbauer, W. (2010). Conducting meta-analyses in R with the metafor package. Journal of

Statistical Software, 36(3), 1-48. Retrieved from http://brieger.esalq.usp.br/CRAN

/web/packages/metafor/vignettes/metafor.pdf

*Walsh, J. A., & Walsh, R. A. (1990). Studies of the maintenance of subsidized foster placements in the Casey Family Program. Child welfare, 69(2), 99-114. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/2107065

*Ward, H., & Skuse, T. (2001). Performance targets and stability of placements for children long looked after away from home. Children & society, 15(5), 333-346. doi:

10.1002/chi.673

*Webster, D., Barth, R. P., & Needell, B. (2000). Placement stability for children in out-of-home care: A longitudinal analysis. Child welfare, 79(5), 614-632. doi:10

4021/2000//050614-19

*Weiner, D. A., Leon, S. C., & Stiehl, M. J. (2011). Demographic, clinical, and geographic predictors of placement disruption among foster care youth receiving wraparound services. Journal of Child and Family Studies, 20(6), 758-770. doi:10.1007/s10826 011-9469-9

(40)

40 Wibbelink, C. J. M., & Assink, M. (2015). Handleiding voor het uitvoeren van een drie-level

meta-analyse in R [Manual for conducting a three-level meta-analysis in R].

Amsterdam: Universiteit van Amsterdam.

Williams, C. & Soydan, H. (2005). When and how does ethnicity matter? A cross-national study of social work responses to ethnicity in child protection cases. British Journal of

Social Work, 35, 901-920. doi: 10.1093/bjsw/bch281

Wilson, K. (2006). Can foster carers help children resolve their emotional and behavioral difficulties? Clinical Child Psychology and Psychiatry, 11, 495-511. doi:10.1177 /1359104506067873

Winokur, M. A., Holtan, A., & Batchelder, K. E. (2014). Matched comparison of children in kinship care and foster care on child welfare outcomes. Families in society, 89, 338 346. doi:10.1606/1044-3894.3759

*Wulczyn, F., Kogan, J., & Harden, B. J. (2003). Placement stability and movement trajectories. Social Service Review, 77(2), 212-236. doi:10.1086/373906

Referenties

GERELATEERDE DOCUMENTEN

The original probe consists of mCitrine (YFP, yellow fluorescent protein) and mCerulean314 (CFP, cyan fluorescent protein), which form a FRET pair, and are connected by a

The lines are calculated intensity ratios for the ion-target combinations from the depth pro file resulting from TRIDYN simulations for a nitrogen incorporation up to

helicopter response pilot controller input actual helicopter dynamics command model analysis of flight test data model validation system ID model inversion model of helicopter

NFR 2 De data flow visualisatie tool moet gebruik kunnen maken van Microsoft SQL server NFR 3 De data flow visualisatie tool moet metadata interpreteren van de ISMetadata database

In Experiment 1, it was investigated how short task sequences of examples and/or practice problems (i.e., EEEE, EPEP, PEPE, and PPPP) would affect motivational (i.e.,

2009 out of IGFA: “a high level group of the world’s major and emerging funders of global environmental change research and international science councils [which] acts

Woestijnvegetatie is hier schaars (groene curve), maar wei rijk aan eiwit (blauwe curve). Schade ont- staat in het droge seizoen, als het vee zich concentreert in de

Gemiddelde waterbalans op jaarbasis De waterbalans van een stroomgebied kan beschreven worden met de volgende formule: Q = P - E - AS + / - Kwel/Wegzijging, waarin, Q p E AS