1
Faculty of Social and Behavioural Sciences
Graduate School of Childhood Development and Education
Using the Strengths and Difficulties
Questionnaire (SDQ) to predict foster
placement breakdown
Research Master Child Development & Education Thesis 1
Dieke Folkeringa
Supervisors: F. van Rooij & B. Zijlstra 15 July 2015
2 Abstract
Foster placement breakdown is considered harmful for the foster child, foster parents and the child welfare system. Identifying predictors might assist in the allocation of services to prevent placement breakdown. Predictors of placement breakdown as identified during previous research are: older age, previous out of home placements, behavior problems, impact on foster parents, and type of placement. The current study aims to provide more information on the relationship between risk factors and placement breakdown. Specifically, it is investigated whether the Strengths and Difficulties Questionnaire (SDQ), which
measures child problem behavior and impact on the foster family, can be considered a reliable tool to predict risk of placement breakdown and whether it is of additional predictive value when background variables are already known. In total, foster families of 361 foster children (aged 4 -12) of two regional foster organizations filled out the SDQ questionnaire. In addition, file analyses were conducted at two foster care organizations to gather placement breakdown data. For 6 cases, placement status was found to be unclear. Cross-tabulations and independent t-tests were conducted to compare two types of placement outcomes: placement continuation (n = 330) versus placement breakdown (n = 25). Logistic regression was used to explore whether subscales of the SDQ and the impact supplement can be considered
predictors for placement breakdown. The results showed that the impact supplement, age upon entering the foster family, and the presence of other foster children at home were significantly positively related to placement breakdown although this only was a small effect. The main model accounted for 30%, the second model accounted for about 29% and the third model accounted for 9% of the variance in foster placement breakdown. These results
indicate that impact may be a better predictor than problem behavior.
Keywords: foster care, placement breakdown, prediction, Strengths and Difficulties Questionnaire, impact
3
Acknowledgements
I would like to express my sincere appreciation for those that have contributed to this research. Thank you Dr. Floor van Rooij for your continuous guidance and for monitoring and encouraging me throughout the writing of this thesis. I am also very grateful for and sincerely appreciate all the valuable feedback and learning opportunities you provided. I would also like to thank Dr. Bonne Zijlstra for providing valuable feedback, and answer all my questions on statistical techniques. I would also like to thank drs. Anne Maaskant for taking the time to read my thesis and provide valuable feedback. Last but not least, I would like to thank the participating foster care organizations for allowing me to conduct file analyses and I would like to thank the staff members for helping me whenever I had questions.
4 Using the Strengths and Difficulties Questionnaire (SDQ)
to predict foster placement breakdown
The number of children in Dutch foster care has increased rapidly over the last decade. In 2013, over 21.600 children under the age of twenty-one were placed in foster care, compared to 10.000 children in 1999. The majority of foster children (69%) were younger than twelve years (Foster care Netherlands, 2013). The Dutch foster care system is comprised of two distinct categories of foster care. The first module of foster care consists of short-term placements that are very often preceded by a family crisis, and are intended to end with family reunification (Foster care Netherlands, 2012). The second module consists of long-term placement of children for whom it might not be possible to return to their biological parents. Continuity, secure upbringing, and optimal developmental opportunities are pursued by placing the child in a foster home where he can stay until adulthood. If possible, contact with biological parents is encouraged (Foster care Netherlands, 2012; Pagée, van Miltenburg, & Pasztor, 1991).
Many foster children have been exposed to numerous adverse circumstances such as neglect, abuse, domestic violence and parental loss. These adverse circumstances may lead to the foster child experiencing insecure attachments and social, emotional, behavioral, and academic problems (Rock, Michelson, Thomson, & Day, 2015). Although foster care aims to improve the living conditions of children who were exposed to inadequate parenting, foster placement breakdown may intensify existing difficulties (Rock, Michelson, Thomson, & Day, 2015). Here, foster placement breakdown refers to unplanned reasons of exiting foster care and are no exceptions in out-of-home care (Chamberlain et al., 2006; Olsson, Egelun, & Høst, 2012; Strijker, Knorth, & Knot-Dickscheit, 2008). According to Strijker, Zandberg and Van der Meulen (2002), an estimated 29% of the foster placements are terminated
5 12.5 and 48% (Holtan, Handegård, Thørnblad, & Vis, 2013; Minty, 1999; Sallnäs,
Vinnerljung, & Westermark, 2004; Strijker et al., 2008; Wilson, Sinclair, & Gibbs; 2000). It is believed that premature foster care termination can be harmful for a child’s development. After foster placement breakdown, foster children need to adapt to a different social and physical environment. With regard to schooling, children who undergo placement changes will be more likely to obtain worse academic results and to suffer from other problems at school. The loss of social relations and educational experiences increases the probability of suffering from emotional and behavior problems. Increased internalizing and externalizing problem behavior of foster children in turn may increase the chance on placement
breakdown. The increased chance on placement breakdown may partly be due to the fact that frequent moving may lead to increased difficulties in forming attachments; and hence also with the child’s substitute caregivers. Also, if a foster child moves frequently, there will be less chance of the child returning to its biological family (Holtan et al., 2013; Leathers, 2006; López López, del Valle, Montserrat, & Bravo, 2011; Newton, Litrownik, & Landsverk, 2000; Rostill-Brookes, Larkin, Toms, & Churchman, 2011; Strijker et al., 2002; Strijker et al., 2008; Van Oijen, 2010).
Placement breakdown also affects foster parents. Taylor and McQuillan (2014) found that foster parents experienced stress, feelings of loss, and were emotionally upset after foster placement breakdown. Foster parents may also experience a sense of failure for not being able to cope with the child and then may decide to stop fostering children altogether (López López et al., 2011; Strijker et al., 2008). Additionally, foster placement breakdown is also considered harmful for the child welfare system. This is because the foster care provider will have to invest time and monetary funds to look for a new placement for the foster child and assist the child and his new foster parents (Strijker et al., 2008; Taylor, & McQuillan, 2014). Since foster placement breakdown has adverse effects on foster children, foster parents and
6 the child welfare system, it is important to identify risk factors and predictors of placement breakdown because it might assist in the allocation of services to prevent placement
breakdown (Fisher, et al., 2011; Strijker et al., 2008; Vanderfaeillie, Holen, Trogh, & Andries, 2012).
A substantive amount of research has been devoted to the identification of risk factors of foster placement breakdown since knowledge of risk factors might indicate areas of
intervention. A meta-analysis by Oosterman, Schuengel, Slot, Bullens and Dorelijers (2007) found that children placed at older age experienced more placement breakdown and children in foster care for reasons of abuse went through more placement breakdowns than children in care for neglect. Children with a background of residential care and children with previous placements in foster care were more likely to experience placement breakdown. Moreover, behavior problems appeared a reasonably robust predictor of placement breakdown
(Oosterman, et al., 2007).
Other researchers have found similar results. Risk factors of placement breakdown were older age, higher number of previous placements, non-kinship placement, residential care as first placement, problematic child behavior (externalizing behavior, attachment problems, hyperactivity problems), longer total time in care, separation from siblings, experiencing multiple social workers, non-Dutch background, experience of physical abuse, being placed out of area of origin and poor integration into foster family (Strijker et al., 2002; Strijker et al., 2008; Rock et al., 2015; Taylor, &McQuillan, 2014; Van Rooij, Maaskant, Weijers, Weijers , & Hermanns , 2015). Taylor and McQuillan (2014) found contact with biological family, unsuccessful attempts to sustain the placement, and risk to foster family to be the most significant issues that lead to placement breakdown. Presence of biological children at the foster has also been found to be associated with foster placement breakdown (Rostill-Brookes, et al., 2011; Wilson, et al., 2000). In contrast with the results found by Rock
7 and colleagues (2015), Van Rooij et al. (2015) found that there were significantly more kinship placements in the placements that terminated unplanned.
However, whereas many researchers investigated foster child-related risk factors, it is important to keep in mind that some foster parent-related factors may also be associated with placement breakdown. Farmer, Lipscombe and Moyers (2005) found that foster caregivers overloaded with a high number of stressful life events prior to the foster youth’s arrival experienced more placement breakdowns than less stressed caregivers. Some factors increasing caregiver strain during placement were difficulties in contacting social workers, contact difficulties with the foster children’s biological families, and behavioral problems of the foster child such as conduct problems, hyperactivity and violent behavior (Farmer, et al., 2005). Van Rooij et al. (2015) found parenting stress, as indicated by how often the foster parents found parenting the foster child hard and by the frequency of conflicts between foster child and foster parent, to be associated with placement breakdown. The researchers stated that parenting stress might result from child problem behavior, and may in turn lead to less effective parenting, which might result in more child problem behavior. This can result in a negative cycle.
Although both foster child-related factors and foster parent-related factors are
associated with foster placement breakdown, problem behavior of the foster child seems to be one of the major risk factors associated with foster placement breakdown. If instruments measuring child behavioral problems can identify predictors of placement breakdown, these instruments may also be used as screening instruments in order to offer timely support when risk factors for placement breakdown are present. Some examples of instruments that have been used to identify predictors of foster placement breakdown are the Parent Daily Report Checklist (PDR) (Chamberlain, & Reid, 1987) and a shortened version of the Child Behavior Checklist, as developed by Barber and Delfabbro (2002).
8 The Parent Daily Report Checklist (PDR), focusing on problematic child behavior, has been used to identify predictors of foster placement breakdown (30 items, behavior observation, α = .84, inter-call correlation between 3 baseline calls: r = .64). This checklist is conducted by telephone twice a week for a total of 10 phone contacts. Parents indicate which, if at all, events occurred over the past 24 hours at the time of the phone call (e.g.
aggressiveness, arguing, bedwetting, competitiveness) (Chamberlain, & Reid, 1987). In research by Chamberlain and colleagues (2006) the number of problem behaviors turned out to be related to the foster child’s risk of foster placement breakdown. Foster children with more than 6 problem behaviors had an increased risk of foster placement breakdown. There was also increased risk of foster placement breakdown as the number of children in the home were higher (Chamberlain et al., 2006). Other researchers using the Parent Daily Report Checklist also found a threshold effect of problem behaviors. Namely, regular foster care children with 5 or fewer problem behaviors were at low risk for placement breakdown, but their risk increased 10% for each additional behavior (Fisher, Stoolmiller, Mannering, Takahashi, & Chamberlain, 2011). Moreover, Hurlburt and colleagues (2010) found that placement breakdown occurred more frequently for children with elevated PDR ratings and children living with non-relative foster parents. Good prediction of placement breakdown was achieved with 3-5 weeks of PDR information (Sensitivity was .62 and specificity was .70) (Hurlburt, Chamberlain, DeGarmo, Zhang, & Price, 2010).
The shortened version of the child behavior checklist (8 items, behavior observations, α = .79 on admission, α =.75 on second measurement occasion; Barber, & Delfabbro, 2002) was used by Strijker and colleagues (2008) to predict placement breakdown and was used by Van Rooij et al. (2015) to code problem behavior by investigating case file reports. The questionnaire codes child behavioral behavior problems as reported by foster care workers observations. Older age, more serious behavioral problems and a long placement history of
9 the foster child were found to be predictors of placement breakdown (Strijker et al., 2008).
Much research has been conducted on the relationship between child problem behavior and foster placement breakdown and various instruments have been designed to measure child behavior and to investigate risk factors. However, since the PDR protocol consists of several phone-calls, it is a time-consuming instrument to use; this may be a disadvantage of the instrument. Both the PDR as well as the checklist by Barber and
Delfabbro (2002) only focus on problematic child behavior. As explained above, foster care also impacts the foster family and the impact the foster placement has on the foster family may influence the placement outcome. Since not much research has been conducted on the relationship of predictive value of impact on the foster family and whether or not foster placements are terminated prematurely, more research is still needed on this topic. A measure that contains an impact supplement, in addition to items on problem behavior, is the Strengths and Difficulties Questionnaire (SDQ) (Goodman, & Goodman, 2009). This instrument is widely used, also in the Netherlands, to measure behavioral problems; it is also used for regular behavioral screening within youth- and school health care (Lehmann, Heiervang, Havik, & Havik, 2014; Municipal Health Care Netherlands (GGD), 2006). An advantage of the SDQ is its conciseness as it takes about 5 minutes to fill out the questionnaire, consisting of 25 items of child problem behavior and prosocial behavior, and four items of the impact supplement; hence it is not very invasive for foster parents. Since the impact supplement is the shortest part of the SDQ, it takes the least time for foster parents to fill out.
The psychometric properties and validity of SDQ have been shown to be good in a number of countries, including the Netherlands (Muris, et al., 2003; van Widenfelt, Goedhart, Treffers, & Goodman, 2003). The concurrent validity, internal consistency, inter-rater
agreement and test-retest reliability of the various SDQ subscales were satisfactory for the parent version (Muris, et al., 2003; Stone, Otten, Engels, Vermulst, & Janssens, 2010). Muris
10 and colleagues (2003) found test-retest stability over a two-month interval with almost all intra-class correlations higher than 0.70, except for the prosocial behavior self-report scale (ICC= 0.59). These results were replicated by Achenbach and colleagues (2008) who found that the levels of reliabilities varied between .50 and .70 for the five subscales, between .70 and .80 for the total difficulties scale. No level of reliability was given for the impact supplement.
The current study investigates whether scales of the Strengths and Difficulties Questionnaire and its impact supplement are of additional predictive value for placement breakdown when child background variables are already known. The research questions of this study are: What is the relationship between problem behavior and impact on the foster family, and whether or not foster placements are terminated prematurely? What are the indicated reasons for placement breakdown? Is the Strength and Difficulty Questionnaire an adequate tool to predict risk of placement breakdown and is it is of additional predictive value when child background variables are already known?
This study is of importance to the practical field since it can be investigated whether existing information is sufficient in predicting placement breakdown or whether SDQ items are of added value. If the SDQ turns out to be an adequate predictor of foster placement breakdown, the instrument may be used as a screening measure by foster care workers in order to offer timely support to prevent foster placement breakdown. Adding of the impact supplement may reveal that impact of the placement on the foster family might be related to foster placement breakdown. As explained above, previous studies have shown many risk factors are associated with foster placement breakdown. Taking previous literature research into account more placement breakdown is expected for foster children who have one or a combination of the following risk factors: higher number of previous out-of-home
11 live with stressed foster parents, biological children at the foster home, or high impact on the foster family. It is also expected that the SDQ scales and especially the impact supplement are significantly related to foster placement breakdown and may be of additive value to background variables that are already know when predicting foster placement breakdown.
Method
Procedure
The study received ethical approval from the Ethical Committee of the Research Institute of Child Development and Education of the University of Amsterdam. Two Dutch regional foster care institutions participated.
Two types of data were used to investigate long-term foster placements of foster children aged 4 to 12. First, existing questionnaire data of foster families from two Dutch regional foster care institutions were used (see also Maaskant, van Rooij, & Hermanns, 2014). Data collection for the current study continued after the article was published by Maaskant and colleagues (2014) and was hence more extensive. The questionnaire data were gathered between 2010 and 2014 and were used to screen foster families for participating in a randomized controlled trial of an intervention program for foster children with behavioral problems. The questionnaire data were either filled out by one or both foster parents.
For the current study, the foster parents of 655 foster children received a questionnaire by post, the parents of 439 children responded (67%). The final dataset consisted of 576 foster parents (330 female, 246 male) of 361 children (excluding duplicates of child ID’s, cases with non-heterosexual foster parents, and cases where foster parents did not expect the placement to be longer than a year). Time between SDQ and file analyses differed per case.
In addition to the questionnaire data, file information (archival data) was collected in May 2014 at the foster care organizations in order to research the incidence and reason of foster placement breakdown. For this, a list was collected of names of the children whose
12 foster parents filled out the questionnaire data previously. For privacy reasons, the list was extinguished as soon as the required file information was gathered.
There were two researchers who coded reasons of placement breakdown information as gathered during file analyses. Categories of reasons of placement breakdown were: Foster child went back home, foster child went to an institution for treatment purposes, the
upbringing became too hard due to the foster child’s problem behavior, the foster child caused problems for other children at home, there were problems in the foster family (e.g. divorce, illness), the foster parents experienced problematic relationship with foster care workers or with biological parents. It was decided to use those categories since they came up in previous studies on reasons of placement breakdown (Brown, & Bednar, 2006; Norgate, Warhust, Osborne, Traill, & Hayden, 2012; Taylor, &McQuillan, 2014). If the information did not match any of these categories, the case was coded as other reason. Intercoder
reliability was calculated for all disrupted placements (n = 25; 6.9% of all files) to determine consistency among coders. The intercoder reliability was found to be substantial (Cohen’s Kappa 0.69 (p <.05), 95% CI (0.482, 0.906) (Landis, & Koch, 1977). Afterwards the two researchers discussed the vague cases and agreed upon the final coding of the cases.
Participants
The mean age of the children was 7.52 years (SD = 2.38, min 4.00 – max 11.96); 50.2% were boys and 63.4% of the children were of Dutch origin. Additionally, the mean age of entering the foster family was 3.33 years (SD = 2.94, min 0.00 - max 10.88) and the mean duration of the current placement was 4.03 years (SD = 2.79, min .15 - max 11.30). For 69.5% of the children the current placement was not their first foster placement. The mean number of previous foster care placements was 0.98 (SD = 0.88, min 0.00 - max 4.00).
The mean age of the foster parents was 49.37 years (SD = 9.31, min 22.70 - max 75.62) and 68.1 % were female. The mean length of foster experiences of foster parents was
13 7.04 years (SD = 7.22, min .06 - 40.87 max). In 85.2% of the cases, other children were present at the foster family (in 47.3% other foster children, in 68.3% biological children of foster parents). About a third of the placements (35.8%) were kinship placements.
Measurements
Two types of data were used, existing questionnaire data and file analyses. Questionnaire data
The questionnaire included questions concerning descriptive socio-demographic variables and potential risk factors with regard to foster placement breakdown, and the Strengths and Difficulties Questionnaire (SDQ; Goodman, & Goodman, 2009). The potential child risk factors that were measured were current age, age upon entering the foster family, gender, cultural background (Dutch versus non-Dutch), number of previous foster
placements, and duration of current placement. With regard to the foster family the following characteristics were measured: age of foster parents, gender of foster parent, type of foster family (kinship versus non-kinship), length of foster carer experience, and presence of other children (foster and biological).
The Dutch parental version of the SDQ was used to screen for behavioral and emotional health problems with foster children (Goodman, & Goodman, 2009; Muris, Meesters, & van den Berg, 2003). The SDQ is comprised of 25 items describing child and adolescent attributes relating to five subscales (emotional symptoms, conduct problems, hyperactivity, peer problems, prosocial behavior) that each contain 5 items. The SDQ also contains an impact supplement on the severity of the perceived problems. The Cronbach’s alpha’s of the SDQ subscales in the current study ranged from .71 (peer problems) to .86 (hyperactivity). This indicates that subscales in the current study were reliable. Each item is scored on a three-point scale (0 = ‘not true’, 1 = ‘somewhat true’, and 2 = ‘certainly true’). An example of an item on the emotional symptoms subscale is: ‘My child often complains of
14 headaches, stomach-aches or sickness’. An example of an item on the conduct problems subscale is: ‘My child often lies or cheats’. An example of an item on the hyperactivity scale is: My child is ‘restless, overactive, cannot stay still for long’. An example of an item on the peer problems scale is: ‘My child is picked on or bullied by other children’. Lastly, an
example of an SDQ item on the pro-social behavior scale is: ‘My child is considerate of other people’s feelings’ (Goodman, & Goodman, 2009). Subscale scores of the SDQ were
computed by summing scores on relevant items (after recoding reversed items; range 0-10). Higher scores reflect more difficulties except for the pro-social behavior scale where higher scores indicated less difficulties. A total difficulties score (TDS) was calculated by summing the scores on four subscales: emotional symptoms, hyperactivity-inattention, conduct
problems and peer problems (Muris, et al., 2003) (range 0-40, α = .77).
The impact supplement of the SDQ inquires the respondents about the severity of the perceived problems and about chronicity, distress, social impairment and burden for the family. The four items are scored on a three-point scale: 0 = not at all/only a little, 1 = quite a lot, 2 = a great deal. An example item of the impact supplement is: ‘Do the difficulties put a burden on you or the family as a whole?’ Aggregation of the distress and social impairment scores led to the impact score (Goodman, 1999). The Cronbach’s alpha’s of the impact supplement in the current study was found to be .79, indicating to be reliable.
File analyses
In addition to the SDQ, data were extended with file analyses conducted in May 2014 at the same Dutch regional foster care organizations. Data were collected on whether the foster children, whose foster parents previously filled out questionnaire data, were still in the same foster family at the time of file analyses. If this was not the case, the indicated reason for placement breakdown and the date of departure were reported.
15
Analyses
In many cases, there were multiple sources of information (mothers, fathers) regarding the same child. By using paired t-tests and correlations it was examined whether foster mothers and fosters fathers differed significantly on the SDQ scales and the impact supplement. Preliminary analyses showed that there was a significant difference in the total problem behavior scores between foster mothers (M=11.80, SD=7.55) and foster fathers (M=12.36, SD=7.79), t(224)=2.13, p = .034. Additionally, the answers from foster mothers (M=7.46, SD=2.20) and foster fathers (M=7.14, SD=2.31) on the prosocial behavior differed significantly, t(227)=-2.92, p = .004. The answers of foster mothers and foster fathers on the impact supplement were highly correlated (r = .79) and foster mothers (M=1.40, SD=1.95) and foster fathers (M=1.52, SD=2.13) did not differ significantly on the impact supplement, t(222)=-1.42, p = .158. Although foster parents filled out significantly different answers on the SDQ total difficulties and pro-social scales their answers were highly correlated (r = .86 for total difficulties score, r = .75 for prosocial behavior scale. For that reason, it was decided to merge the answers of both foster parents (if applicable). Since foster mothers and fathers filled out significantly different answers on two SDQ scales, it was checked whether effects on these scales (total difficulties scale and pro-social behavior) were different for mothers and fathers separately.
In preliminary analyses the Fisher’s Exact Test was used to test associations between foster placement breakdown and the categorical variables: gender, type of placement (kinship versus non-kinship), presence of biological children at home, presence of foster children at home, cultural background, and multiple foster placements. Additionally, independent t tests were conducted in order to test associations between foster placement breakdown and the following continuous variables: duration of current placement, age of the foster child, age of foster child at start of placement, number of previous foster placements, age of foster parent,
16 years of foster parent experience, time since SDQ, the SDQ total difficulties scale and the SDQ subscales emotional symptoms, conduct problems, hyperactivity, peer problems, prosocial behavior, and the SDQ impact supplement. Time since SDQ was controlled for if the relationship between time since SDQ and placement breakdown was found to be significant.
For significant variables from the cross-tabulations and independent t tests, a standard logistic regression analysis was conducted to investigate whether the Strength and Difficulty Questionnaire scales in combination with background variables (independent variables) can be used to predict risk of placement breakdown (dependent variables) and whether it is of additional predictive value when child background variables are already known.
Results
Descriptive statistics showed that out of 361 total placements, 330 placements continued at the time of file analyses. For 6 cases it was unclear whether the children were still in the same foster family or not. For 3 of these cases, the children moved to a different province and were supervised by a different foster care organization. Hence at the time of file analyses it was unclear whether these children still lived in the same foster family.
Additionally, there was no information on placement status for the other 3 cases. For that reason the 6 cases were labelled as missing values and were not included in additional
analyses (n = 355). File analyses of the cases of placement breakdown (n =25) indicated that behavioral problems of foster child (32%) was the most indicated reason for foster placement breakdown. In 6 cases (24%) the foster child could return home, and in 4 cases (16%) foster placement breakdown occurred due to circumstances in the foster family such as illness or divorce. In 3 cases (12%) the foster placement broke down because the foster child caused problems for other children in the foster family.
17 2.47 years (SD = 0.74, min .55 - max 3.28) since filling out SDQ information. The mean age of the children at the point of placement breakdown was 9.22 years (SD = 2.77, min 4.89 - max 14.94), and 52% were boys and 48% were girls. The mean duration of the foster placement before breakdown was 2.80 years (SD = 2.60, min 0.16 - max 10.39). From the placements that broke down, 36% were kinship placements. See Table 1 and 2 for descriptive information as calculated for children whose placements had broken down versus placements that had not broken down at time of file analyses.
When looking at the differences between foster mothers and fathers, no significant differences were found on the total difficulties scale. On the pro-social scale, the effects for fathers were not significant, t(244)=0.99, p = .324, the effect for mothers was significant, t(329)=2.72, p = .047. For the merged family scales, the effects of the total difficulties scale and pro-social behavior scales were significant and were hence included in the analyses.
Table 1
Descriptive statistics continuation versus placement breakdown Continuation n = 330
Breakdown n = 25
t DF p
Sex of foster child n (%) 175 (53.0%a) 12 (48% a) . e . e .390
Type of placement n (%) 110 (39.1%b) 11 (36%b) . e . e .592
Presence biological children n (%) 235 (71.2%) 22 (88%) . e . e .055
Presence foster children n (%) 159 (48.2%) 4 (16%) . e . e .001*
Cultural background n (%) 201 (60.9%c) 21 (84% c) . e . e .015*
Previous placements n (%) 229 (69.4% d) 19 (76.0% d) . e . e .328
Sex foster parent n (%) 226 (68.5%a) 18 (72% a) . e . e .453
Duration foster family M (SD) 4.16 (2.84) 2.80 (2.60) 2.33 353 .020* Age foster child M (SD) 7.45 (2.43) 7.98 (2.41) -1.04 353 .297 Age when entering foster family M (SD) 3.24 (2.82) 5.49 (2.82) -3.67 336 .000* Nr. of previous placements M (SD) 0.96 (0.86) 1.16 (1.03) -1.10 353 .271 Age foster parent M (SD) 49.23 (9.18) 49.33 (9.64) -0.05 351 .956 Foster care experience (years) M (SD) 6.96 (6.89) 4.71 (6.05) 1.59 349 .113 Years since SDQ M (SD) 2.26 (0.92) 2.47 (0.76) -1.36 29.65 .185
Note: a: female; b: kinship placement; c: Dutch cultural background; d: yes; e: no value due to the use of Fisher’s Exact test; * p < .05
18 When looking at the relation between risk factors and the dependent variable foster placement breakdown, a significant relationship was found between presence of other foster children at home and placement breakdown (see Table 1). Other foster children were present in a lower percentage in the families where placement broke down prematurely in
comparison with the families where placement did not break down prematurely. Moreover, the relationship between cultural background and placement breakdown was significant. Namely, placements of children with a Dutch cultural background were more likely to break down than those of children with non-Dutch cultural backgrounds. Additionally, the
relationship between age upon entering the foster family and placement breakdown was significant. Placements were more likely to break down if the child entered the foster family at a later age. Lastly, the relationship between placement duration and placement breakdown was significant. Placements of shorter duration were more likely to break down. No
significant difference was found between placement breakdown and time since SDQ. There were no significant differences found between the other variables with regard to foster placement breakdown.
Table 2
Descriptive statistics SDQ scales for continuation versus placement breakdown Continuation n = 330 Breakdown n = 25 M SD M SD t DF p SDQ scale Conduct problems 2.17 2.01 3.52 2.51 -2.63 26.40 .014* Emotional symptoms 2.39 2.25 3.36 2.16 -2.07 352 .039* Hyperactivity 4.84 2.96 6.70 2.51 -3.05 352 .002** Peer problems 2.04 2.03 3.66 2.43 -3.79 352 .000*** Total problem behavior 11.44 7.15 17.24 7.46 -3.90 352 .000*** Pro-social behavior 7.47 2.06 6.38 2.37 2.52 352 .012* Impact 1.26 1.86 2.92 2.67 -3.05 25.81 .005**
19 Concerning the SDQ subscales, significant positive relationships were found between all behavior problem subscales (peer problems, hyperactivity, conduct problems, emotional symptoms) and placement breakdown (See Table 2). The more child behavioral problems foster parents indicated, the more likely it was that the placement broke down.
In addition, a significant negative relationship was found between prosocial behavior of foster children and whether the child still lived in the respective family. The more child prosocial behavior foster parents indicated, the less likely it was that the placement ended prematurely.Impact on the foster family was also found to be positively correlated with placement breakdown and places that had more impact on the foster family were more likely to break down.
A standard logistic regression analyses was performed to examine whether the Strength and Difficulty Questionnaire scales are of additive value in predicting foster placement breakdown if certain background characteristics were already known. Three different models were run. For the main model, it was decided to use several blocks of predictors in the logistic regression in order to research whether SDQ scales were of additive value in predicting foster placement breakdown if certain background characteristics were already known. Concerning SDQ problem behavior scales, only the total problem behavior scales was included in the analyses since it consisted of all SDQ problem behavior scales (conduct problems, emotional problem behavior, hyperactivity, and peer problems). The first block of the model comprised the following demographic variables: cultural background of the foster child, duration in foster family, age upon entering the foster family and presence of other foster children. The second block comprised the impact supplement, total difficulties score and prosocial behavior scale of the Strengths and Difficulties Questionnaire, and was added to the model containing the first block. Table 3 shows how well the model classifies the cases.
20 A test of the main model only containing demographic predictors (block 1) against the constant model was statistically significant, χ2 (4) = 21.213, p < .000, indicating that the predictors significantly distinguished between placement breakdown. Afterwards, a second block, containing SDQ scales, was added to the model. The significant chi-square in block two of the model indicates that the full model significantly predicted foster placement
breakdown, χ2 (7) =41.961, p <.000. The Hosmer and Lemeshow test of the goodness of fit of the second block also suggests the model is a good fit to the data as the p-value was not significant, χ2=5.267, df = 8, p = .729. With the use of the demographic variables in the first block of the main model, 65.2% of the foster children were correctly identified as being at risk of placement breakdown and 80.1% of the foster children were correctly identified as not being at risk of placement breakdown (overall success rate: 79%). With the use of the SDQ variables in the second block, 52.2% of the foster children were correctly identified as being at risk of placement breakdown and 83.9% of the foster children were correctly identified as not being at risk of placement breakdown. The first block of the main model accounted for 17% of the variance in foster placement breakdown and the final model, including block 2, accounted for 30% of the variance in foster placement breakdown. Even though the second block of the full model identified fewer children at risk of placement breakdown, more variance in foster placement breakdown was explained by adding SDQ scales to the background variables that are already known.
Table 3 shows the regression coefficients, odds ratios and 95% confidence intervals for odds rations for each of the predictors off the full model. Significant Wald Chi -square values demonstrated that the presence of other foster kids at home, the age upon entering the foster family, and the impact scale of the SDQ made significant contributions to the
prediction of foster placement breakdown prediction. The value of the coefficient of presence of other foster children at home revealed that, given the other predictors in the model,
21 placements where no other foster siblings lived at home were 3.66 times more likely to be terminated unplanned compared to foster families where other foster children lived at home. It was also found that, given the other predictors in the model, the placements of foster children that were older when entering the foster family were 1.25 times more likely to break down in comparison with the placements of foster children that were younger when entering the foster family. Lastly, given the other predictors in the model it was found that higher answers on the impact scale were associated with an increase in the odds of unplanned termination by a factor of 1.42.
Table 3
Logistic regression for model including demographics and SDQ scales (n = 355)
95% CI for odds ratio Variables B S.E. Wald
Chi-square p Lower Odds ratio Upper Main model Constant -6.88 1.74 15.69 .000* .001 Block 1 Child characteristics
Culture foster child a 0.76 .61 1.55 .213 .65 .14 7.308 Age upon entering the foster family 0.22 .11 4.07 .044* 1.01 1.25 1.55 Placement characteristics
Presence other foster children a 1.30 .62 4.34 .037* 1.08 3.66 12.38 Duration placement a -0.03 .13 .068 .794 .75 .97 1.24
Block 2
SDQ scales
Total problem behavior 0.06 .05 1.38 .240 .96 1.06 1.16 Prosocial behavior 0.06 .13 .18 .672 .81 1.06 1.38 Impact supplement 0.35 .14 6.00 .014* 1.07 1.42 1.87
Model 2
Constant -6.13 1.23 24.86 .000* .002 Child characteristics
Culture foster child a 0.89 .61 2.14 .144 .74 2.43 8.01 Age upon entering the foster family 0.23 .11 4.80 .028* 1.02 1.26 1.56 Placement characteristics
Presence other foster children a 1.35 .63 4.68 .030* 1.14 3.87 13.18 Duration placement a -0.01 .13 0.00 .954 .77 .99 1.27 Impact scale 0.45 .10 18.90 .000* 1.28 1.57 1.92
Model 3
Constant -3.22 .31 104.92 .000* .040
Impact scale 0.33 .09 14.28 .000* 1.17 1.39 1.64
Note: Main model: First block: Nagelkerke R2: .17; χ2 (4) = 21.213, df=4, p < .000; Second block: Nagelkerke R2: .30; χ2 (7) =41.961, df=7, p <.000; Model 2: First block: Nagelkerke R2: .16; χ2(4) = 21.213, df=4, p < .01; Second block: Nagelkerke R2: .29; χ2 (5) = 40.586, p < .01; Model 3: Nagelkerke R2: .09; χ2(1) = 13.464, df=1, *p < .05, a = dichotomous variable
22 A second model (Model 2) was run, and it was also decided to use two blocks of predictors in the logistic regression analysis to research whether only the impact supplement of the SDQ was of additive value in predicting foster placement breakdown if certain
background characteristics were already known. The first block of model 2 consisted of the demographic variables cultural background, presence of foster children at home, age upon entering the foster family, and duration of foster placement. The second block of model 2 consisted of the impact scale of the SDQ (See table 3). A test of the full model 2 with the impact scale predictor against a constant model only was statistically significant, χ2 (5) = 40.586, p < .01. The Hosmer and Lemeshow test of the goodness of fit also suggests the model is a good fit to the data as the p-value was not significant, χ2 (8) =10.004, p = .265. The model (model 2) accounted for about 29% of the variance in foster placement
breakdown. With the use of the demographic variables in the first block, 65.2% of the foster children were correctly identified as being at risk of placement breakdown and 80.1% of the foster children were correctly identified as not being at risk of placement breakdown. With the use of the impact scale in the second block of model 2, 47.8% of the foster children were correctly identified as being at risk of placement breakdown and 83.3% of the foster children were correctly identified as not being at risk of placement breakdown (overall success rate: 80.8%).
Additionally, a third model (model 3) was run with only the impact supplement of the SDQ included as an independent variable (see Table 3) to investigate whether the impact on the foster family could predict foster placement disruption (dependent variable). A test of model 3 with the impact scale predictor against a constant model only was statistically significant, χ2 (1) = 13.464, p < .01. The Hosmer and Lemeshow test of the goodness of fit also suggests the model is a good fit to the data as the p-value was not significant, χ2 =1.858, df=3, p = .602. The model (model 3) accounted for about 10% of the variance in foster
23 placement breakdown with correct prediction of 48% of the cases of placement breakdown and correct prediction of 83.7% of the cases that did not breakdown (overall success rate: 81.2%).
Thus, both the main model (including background variables and SDQ scales and impact supplement) as well as the model including only background variables and impact supplement account for about 30% of the variance in foster placement breakdown, and the model including only the impact supplement account for 9% of the variance in foster placement breakdown. The results show that more explained variance in foster placement breakdown was accounted for by adding SDQ scales to the background variables that are already known. Results also showed that with the use of the SDQ variables, the main model (model 1) as well as model 2 and model 3 were able to correctly predict about half of the cases of placement breakdown (47.8% - 52.2%) and over 83% (83.3% - 83.9%) of the cases of placement continuation. This percentage is lower than before SDQ scales or the impact supplement were added in model 1 and model 2, since 65.2% of cases placement breakdown and 80.1% of placement continuation were correctly predicted by only using child
background characteristics.
Discussion
The current study investigated the relationship between problem behavior and impact on the foster family, and whether or not foster placements are terminated prematurely. Moreover, indicated reasons for placement breakdown were investigated by means of file analyses. Lastly, it was aimed to find out by running logistic regression models whether scales of the Strength and Difficulty Questionnaire predicted risk of placement breakdown and whether the SDQ is of additive value when background characteristics are already known.
Whereas previous research indicated that placements of older foster children were more likely to break down, interestingly enough this was not the case in the current study.
24 Placements of foster children that experienced previous placements were also not more likely to break down. In contrast with previous research, non-kinship placements were not more likely to break down. Whereas it was expected that shorter duration of the foster placement would be positively related to foster placement breakdown, this was not found in the current study. Also contrary to previous research, presence of biological children was not found to be significantly positively related to foster placement breakdown (Barber, & Delfabbo, 2002; Oosterman et al., 2007; Rock et al., 2015; Rostill-Brookes, et al., 2011; Strijker et al., 2002; Strijker et al., 2008; Rock et al., 2015; Taylor, &McQuillan, 2014; Van Rooij, Maaskant, Weijers, Weijers & Hermanns , 2015; Wilson, et al., 2000).
However, several significant relationships were found between foster placement breakdown and various risk factors: presence of other foster children at home, cultural background, age upon entering the foster family, and placement duration. Specifically, in significantly more cases of foster breakdown, no other foster children were present. Also, for the placements that broke down, there were significantly more children of Dutch origin and the foster child had entered the foster family at a later age. Although the current study only investigated long-term placements (placement duration intention of > 1 year), it was found that in comparison to placements that did not break down, placements that did break down were of shorter duration. This is in line with most previous research. However, van Rooij et al (2015) found no significant difference between duration of placement (short term versus long-term) and placement breakdown.
In addition to the background variables, significant relationships were found between foster placement breakdown and all SDQ subscales as well as the impact supplement of the SDQ. In line with previous research, the more child behavioral problems were indicated by foster parents, the more likely it was that placements broke down (Oosterman, et al., 2007; Rock, et al., 2015). As expected from previous research indicating that placements affect
25 foster parents (López López et al., 2011; Strijker et al., 2008; Van Rooij et al., 2015),
placements were more likely to break down if they impacted the foster family significantly. Lastly, the more prosocial behavior foster parents indicated the less likely it was that the placement ended prematurely. Also in line with previous research were the reasons of
placement breakdown. File analyses revealed that the reasons of placement breakdown in the current study (behavioral problems of foster child, the foster child could return home, to circumstances in the foster family, caused problems for other children in the foster family overlapped with previous studies (Brown, & Bednar, 2006; Norgate, Warhust, Osborne, Traill, & Hayden, 2012; Taylor, &McQuillan, 2014).
A significant positive relationship was found between problem behavior, impact on the foster family, and placement breakdown, as measured by the SDQ. Because all SDQ scales as well as its impact supplement were significantly related to foster placement
breakdown, the Strength and Difficulty Questionnaire may be used as a tool to predict risk of placement breakdown. Logistic regression was performed to examine whether the SDQ is of additional predictive value to child background variables.
The current study revealed some interesting findings. Results showed that with the use of the SDQ variables, all three models were able to correctly predict about half of the cases of placement breakdown and over 83% of the cases of placement continuation.
Interestingly, this percentage is lower than before SDQ scales or the impact supplement were added in model 1 and model 2. Thus, against our expectations, the SDQ does not seem to be of additive value to background variables that are already known in predicting foster
placement breakdown. The original model (Model 1) as well as model 2 accounted for about the same percentage of explained variance (30%) and correctly predicted the same percentage of cases of placement breakdown (about 50%). This indicates that it may be possible to only have foster parents fill out the impact supplement of the SDQ to get the same predictive value
26 for foster placement breakdown as when all SDQ scales and the impact supplement are included. The finding that the model (model 3) including only the impact supplement, was able to predict about the same percentage of cases of placement breakdown and placement continuation indicates that no knowledge of child background characteristics are needed to be able to correctly predict about half of the cases of foster breakdown and over 83% of the cases of foster continuation.
The placement breakdown rate in the present study (6.9%) was much lower than what is found in other studies (ranging between 12.5% - 48%). This might be due to the age span, since relatively young children (primary school age) were included in the present study. As found in many studies, placements of older children are more likely to break down (Barber, & Delfabbo, 2002; Oosterman et al., 2007; Rock et al., 2015; Strijker et al., 2002; Strijker et al., 2008). Another reason why placement breakdown might be lower than expected is due to the time-frame, because in some cases the file analyses were conducted shortly after SDQ information was filled out by foster parents. Hence more cases of placement breakdown may have appeared in case there was more time between questionnaires and file analyses.
Nevertheless, no significant relationship was found between time since SDQ and foster placement breakdown. Another explanation for the relatively low percentage of placement breakdown in comparison with other studies could be the type of foster placement module. This is because short-term placements are more likely to break down than long-term
placements, and the current study consists only of (intended) long-term foster placements (> one year) (Oosterman et al., 2007; Strijker et al., 2002; Strijker et al., 2008). A reason why short-term placements are more likely to break down is because there is less time to develop a bond between the foster child and its foster parent(s). Especially for children with insecure attachments, such as foster children, it may be a challenge and take time to adapt to a new context and caretakers (Tucker, & Mackenzie, 2012).
27 The current study found that it was more likely for placements to break down if no other foster siblings were present at home. A possible explanation could be that in these cases there was no possibility to look for distraction by shifting attention to other foster children. Specifically, as people become more stressed out, attention tends to focus on what is perceived as distressing. This process is called attentional narrowing (Kuczynski, 2002). Attention diversion is one of the ways for stress relief, and this may be required in difficult or stressful situations that can neither be avoided nor altered substantially (Jaremko &
Meichenbaum, 2013). Hence, the presence of other foster children at home may help disrupt the process of attentional narrowing by providing an opportunity for foster parents to shift their attention. This may then contribute to stress reduction and give foster parents the possibility to deal better with the demanding (situation of the) foster child.
The result that placements of children with a Dutch cultural background were more likely to break down than those of children with mixed cultural backgrounds may be due to cultural identity. Whereas in many Western cultures, there is a belief of the independent self and the focus is on the individual well-being, in many non-western cultures, there is a more interdependent, group-oriented outlook and a larger on the well-being of the group (Aronson, Wilson, & Akert, 2010). Consequently, it may be that in non-Dutch foster families, there is more emphasis on the well-being of other people in the family (e.g. foster child) instead of on your own well-being (as foster parent).
There were several limitations to this study. A major limitation is the sample size of the cases of placement breakdown (n = 25); this should be taken into account when
interpreting the results. Secondly, since there were several waves (from February 2011 until February 2014) for parents filling out the SDQ, time since SDQ was not equivalent for all cases. Specifically, at time of file analyses, some parents filled out the questionnaires three years prior to that time, others just filled it out a few months earlier. This is a disadvantage
28 because it is not possible to measure placement breakdown after the same time period (e.g. one year after questionnaire research). Secondly, there were limitations due to the type of information used in this study. All information as retrieved from file analyses, was noted down by foster care workers who were in contact with foster parents. The indicated reasons for placement breakdown might have been differently described by foster parents. Lastly, in some cases, the files did not entail information on whether children were still living in the current family. Hence, more research should be conducted on the predictive value of the SDQ with a larger sample size of cases of placement breakdown. Also, it would be valuable if all foster parents fill out the SDQ after a set amount of time (e.g. 1 year after placement; 2 years after placement) so that the results can be compared for the different time periods.
Notwithstanding these limitations, the present study provides unique information on the additive value of the Strengths and Difficulties Questionnaire to child background variables in predicting foster placement breakdown.
29 References
Achenbach, T., Becker, A., Dopfner, M., Heiervan, E., Roessner, V., Steinhausen, H. C., & Rothenberger, A. (2008). Multicultural assessment of child and adolescent
psychopathology with ASEBA and SDQ instruments research findings, applications, and future directions. Journal of Child Psychology and Psychiatry and Allied
Disciplines, 49(3), 251-275.
Aronson, E., Wilson, R.D., & Akert, R. M. (2010). Social Psychology. Seventh edition. Boston, USA: Pearson Education.
Barber, J. G., & Delfabbro, P. H. (2002). The plight of disruptive children in out-of-home care. Children’s Services, Social Policy, Research, and Practice, 5(3), 201-212. Brown, J. D., & Bednar, L. M. (2006). Foster parent perceptions of placement breakdown.
Children and Youth Services Review, 28(12), 1497-1511.
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.
Chamberlain, P., & Reid, J. B. (1987). Parent observation and report of child symptoms. Behavioral Assessment, 9(1), 97-109.
Crone, M. R., Vogels, A. G. C., Hoekstra, F., Treffers, P. D. A., & Reijneveld, S. A. (2008). A comparison of four scoring methods based on the parent-rated strengths and difficulties questionnaire as used in the Dutch preventive child health care system. BMC Public Health, 8(106), 1-9.
Farmer, E., Lipscombe, J., & Moyers, S. (2005). Foster carer strain and its impact on parenting and placement outcomes for adolescents. British Journal of Social Work, 35(2), 237-253.
30 Foster placement disruptions associated with problem behavior: Mitigating a
threshold effect. Journal of Consulting and Clinical Psychology, 79(4), 481-487. Foster care Netherlands. (2013). Foster Care Fact Sheet. Retrieved from
https://www.pleegzorg.nl/over-pleegzorg/factsheet/.
Goodman, R., (1999). The extended version of the Strengths and Difficulties Questionnaire as a guide to child psychiatric caseness and consequent burden. Journal of Child Psychology and Psychiatry, 40(5), 791-799.
Goodman, A., & Goodman, R. (2009). Strengths and difficulties questionnaire as a dimensional measure of child mental health. Journal of the American Academy of Child & Adolescent Psychiatry, 48(4), 400-403.
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.
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.
Jaremko, M., & Meichenbaum. D. (2013). Stress Reduction and Prevention. New York, USA: Springer Science & Business Media.
Kuczynski, L. (2002). Handbook of Dynamics in Parent-Child Relations. Family & Relationships. Thousand Oaks, CA, USA: SAGE Publications.
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.
Lehmann, S., Heiervang, E. R., Havik, T., & Havik, O. E. (2014). Screening foster children for mental disorders: Properties of the Strengths and Difficulties Questionnaire. PLoS
31 ONE, 9(7), e102134.
Lo, A., Roben, C. K. P., Maier, C., Fabian, K., Shauffer, C., Dozier, M. (2015). “I want to be there when he graduates:” Foster parents show higher levels of commitment than group care provider. Children and Youth Services Review, 51, 95-100.
López López, M., del Valle, Jorge F., Montserrat, M., & Bravo, A. (2011). Factors affecting foster care breakdown in Spain. The Spanish Journal of Psychology, 14(1), 111-122. Maaskant, A. M., van Rooij, F. B., & Hermanns, J. M. A. (2014). Mental health and
associated risks factors of Dutch school aged foster children placed in long-term foster care. Children and Youth Services Review, 44, 207-216.
Minty, B. (1999). Annotation: Outcomes in long-term foster family care. Journal of Child Psychology and Psychiatry, 40(7), 991-999.
Municipal Health Care Netherlands (GGD) (2006). Guidelines for the use of the SDQ in youth healthcare. A questionnaire for the identification of psychosocial problems in children aged 7-12 years. Instruments psychosocial problems (LSPPJ). Retrieved from http://mijn.qind.nl/userfiles/192/File/Handleiding%20SDQ.pdf
Muris, P., Meesters, C., & van den Berg, F. (2003). The strengths and difficulties
questionnaire (SDQ): Further evidence for its reliability and validity in a community sample of Dutch children and adolescents. European Child & Adolescent Psychiatry, 12(1), 1-8.
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 placements. Child Abuse & Neglect, 24(10), 1363-1374.
Norgate, R., Warhust, A. Osborne, C., Traill, M., & Hayden, C. (2012). Social Workers' Perspectives on the Placement Instability of Looked after Children. Adoption & Fostering, 36(2), 4-18.
32 Olsson, M., Egelund, T., & Høst, A. (2012). Breakdown of teenage placements in Danish
out-of-home care. Child and Family Social Work, 17(1), 13-22.
Oosterman, M., Schuengel, C., Slot, N. W., Bullens, R. A. R., & Doreleijers, T. A. H. (2007). Disruptions in foster care: A review and meta-analysis. Children and Youth Services Review, 29(1), 53-76.
Pagée, R., van Miltenburg, W., & Pasztor, E. M. (1991). The international transfer of foster parent selection and preparation technology: The example of the Netherlands and the United States. Child Welfare. 70(2), 219-227.
Palmer, S. E. (1996). Placement stability and inclusive practice in foster care: An empirical study. Children and Youth Services Review, 18(7), 589-601.
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(1), 177-203. Rostill-Brookes, H., Larkin, M., Toms, A., & Churchman, C. (2011). A shared experience of
fragmentation: Making sense of foster placement. Clinical Child Psychology and Psychiatry, 16(1), 103-127.
Sallnäs, M., Vinnerljung, B., & Westermark, P. K. (2004). Breakdown of teenage placements in Swedish foster and residential care. Child and Family Social Work, 9, 141-152. Stone, L. L., Otten, R., Engels, R. C. M. E., Vermulst, A. A., & Janssens, J. M. A. M. (2010).
Psychometric properties of the parent and teacher versions of the strengths and difficulties questionnaire for 4- to 12-year-olds. A review. Clinical Child and Family Psychology Review, 13(3), 254-274.
Strijker, J. (2009). Kennisboek pleegzorg. Groningen: Stili Novi.
Strijker, J., Knorth, E. J., & Knot-Dickscheit, J. (2008). Placement history of foster children: A study of placement history and outcomes in long-term family foster care. Child
33 Welfare, 87(5), 107-124.
Strijker, J., Zandberg, T., & van der Meulen, B. F. (2002). Indicators for placement in foster care. British Journal of Social Work, 32(2), 217-231.
Strijker, J., Zandberg, Tj., & van der Meulen, B.F. (2003). Kinship foster care and foster care in the Netherlands. Children and Youth Services Review, 25(11), 843-862.
Taylor, B. J., & McQuillan, K. (2014). Perspectives of foster parents and social workers on foster placement disruption. Child Care in Practice, 20(2), 232-249.
Tucker, D. J., & MacKenzie, M. J. (2012). Attachment theory and change processes in foster care. Children and Youth Services Review, 34(11), 2208-2219.
Vanderfaeillie, J., Van Holen, F., Trogh, L., & Andries, C. (2012). The impact of foster children’s behavioural problems on Flemish foster mothers’ parenting behavior. Child and Family Social Work, 17(1), 34-42.
Van Montfoort, A. J., & Tilanus C. P. G. (2007). Jeugdzorg en jeugdbeleid. Amsterdam: SWP Publishing Company.
Van Oijen, S. (2010). Resultaat van pleegzorgplaatsingen. Een onderzoek naar breakdown en de ontwikkeling van adolescente pleegkinderen bij langdurige
pleegzorgplaatsingen. [The result of foster care placements. A study on the
breakdown and development of adolescent foster children in long-term foster care.]. PhD Dissertation. Groningen, the Netherlands: Rijksuniversiteit Groningen.
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.
Van Widenfelt, B. M., Goedhart, A. W., Treffers, Ph. D. A., & Goodman, R. (2003). Dutch version of the strengths and difficulties questionnaire (SDQ). European Child &
34 Adolescent Psychiatry, 12(6), 281-289.
Wilson, K., Sinclair, I., & Gibbs, I. (2000). The trouble with foster care: The impact of stressful events on foster carers. British Journal of Social Work, 30(2), 191-209.
