Introduction

Subjective well-being is an essential non-material component in the quality of life of children. It is related to children’s perceptions and evaluations of their main life domains, and their aspirations (Campbell et al. 1976; Diener 2012), studied here from a social science, rather than a health-related perspective. However, little research has focused on the subjective well-being of vulnerable sub-groups, such as children in out-of-home care. We could mention Tomyn (2013) who, in his longitudinal study of at-risk population in Australia, observed that youth in care had significantly lower levels of subjective well-being than the general population of the same age. Girls scored lower in subjective well-being than boys and were more prone to depression. In contrast, youth in care who maintained regular contact with their friends had higher levels of well-being.

Examples of the growing interest in SWB and its application can be found in Selwyn et al. (2016), who has been developing instruments to measure the subjective well-being of children in foster care in the United Kingdom based on a qualitative approach and who, together with Davern (2016) in Melbourne, has backed the use of SWB measurement to assess childhood programs. Yet, the studies that marked the starting point for our research were by Llosada-Gistau et al. (2015, 2017) in Spain; Schütz et al. (2015) in Brazil, addressing only children in residential care, and Rees et al. (2012) in the United Kingdom, on children in foster families. All these studies coincided in identifying lower levels of subjective well-being in children in residential and family foster care than in the general population of the same age. Llosada-Gistau et al. (2017) identified factors leading to lower well-being among the population in care, such as opposition to the out-of-home placement, changing schools, low academic achievement, and being unable to see their friends or take part in extra-curricular activities. They also began to compare children in different kinds of out-of-home placements and showed that children in residential care not only displayed lower levels of subjective well-being compared to the general population, but also to the population in kinship or non-kin foster care.

Regarding children in kinship care, authors differ as to the outcomes of this type of placement, which is highly determined by the context in which it takes place. Taussig and Clyman (2012) showed how their study findings suggested a pattern of poorer functioning for youth who spent more time living in kinship care. Without wishing to enter into a debate, we have focused mainly on research more directly related to the study of well-being. One such example is the study by Montserrat and Casas (2007), in which the authors indicated, albeit tentatively and provisionally, similar levels of self-reported subjective well-being for children in kinship care as the general population. In Canada, Schwartz et al. (2014) also indicated that most children preferred kinship care with a relative over non-kin or other types of out-of-home placements and they highlighted the stability provided by this type of care, since children in kinship care had fewer placement changes, were less likely to have their placement disrupted and their foster parents tended to be more highly committed. This led to them having better results in emotional well-being and less likelihood of having behavioural problems or mental disorders. Authors like Farmer and Moyers (2008) and Burgess et al. (2010) also pointed out that these children tended to maintain contact with the rest of the family and establish a sense of belonging more easily.

In contrast, children in residential care and in non-kinship foster care were often exposed to far greater instability due to placement changes, which had an impact on their schooling, friendships and extra-curricular activities (Rees et al. 2012; Wade et al. 2011) compared to those in kinship care (Montserrat et al. 2015). In addition, González-García et al. (2017) found that 49% of children in residential care in Spain were receiving some kind of mental health treatment and 61% were identified as within the clinical range in some of the broad band scales of the Child Behavior Checklist (CBCL), indicating high levels of difficulties among the residential foster care population, thus re-opening the debate on the therapeutic role of residential out-of-home care (Whittaker et al. 2016). As for the well-being of children in residential care, research findings were not unanimous either, and few studies are known to have focused exclusively on subjective well-being. Llosada-Gistau et al. (2015) pointed out that results achieved in well-being scales by children in residential care were lower than the general population of the same age. This paper goes a step further with a more in-depth analysis of both the stability factor and the child-caregiver relationship, examining in what way these two important domains contribute to children’s well-being in different type of placements, and includes quantitative and qualitative data The fact that this study has been based on the child’s perspective means that children have been considered active subjects with a right to participate in matters that directly affect their lives, and the qualitative approach enriches and allows a better understanding of their views and opinions. However, research for collecting qualitative data aimed at determining the subjective well-being of children in care is still underdeveloped (Selwyin et al. 2016).

Study Objectives

Based on the few studies that have tackled the issue of the subjective well-being of children in out-of-home care, this study was aimed at furthering knowledge about the factors influencing subjective well-being (SWB) from a comparative perspective, comparing the different types of placements. Cross-checking factors at the heart of the child welfare system with subjective well-being scales is a new approach in the context of scientific research. More specifically, this study aimed to analyse the impact that the following factors may have on the subjective well-being of children both in residential and foster care (kinship and non-kinship) in Catalonia (Spain):

  1. (a)

    type of placement, age and gender,

  2. (b)

    placement stability within the care system: length of time in the same placement, number of changes, previous foster care disruptions, number of places in residential homes and number of times the child had run away

  3. (c)

    relationships with caregivers and other children in the same placement.

Factors (b) and (c) are closely linked from a systemic viewpoint given that a certain amount of stability is needed to be able to establish satisfactory relationships. Moreover, these two factors have been counter-balanced with qualitative data, as we felt it was important to understand the situation from the perspective of the children themselves.

Project Setting: Out-Of-Home Care in Catalonia (Spain)

In June 2016, there were 6840 children under 18 years of age in out-of-home placements in Catalonia (5 per 1000 children), mainly due to physical neglect, but also to emotional neglect and psychological, physical and sexual abuse. Forty-one percent were in residential care, 36% in kinship care and 20% in non-kin care. The remaining 4% were in provisional care (DGAIA, Directorate General for Child and Adolescent Care,2016). These figures were similar in the rest of Spain, where 38% were in residential care, 46% in kinship care and 16% in non-kin foster care (MSSSI 2015).

Residential care in Catalonia and in Spain has never been conceived as mainly for children with significant mental health needs. The large percentage of the population in residential settings has several origins and reasons: the historical and cultural tradition, and the lack of foster families, often due to a lack of political will and a preference for keeping siblings together, difficult to achieve in the same foster family. Nonetheless, this does not mean that some children in residential care do not have mental health issues, as pointed out by González-García et al. (2017). In Catalonia, only 1 of the 146 residential foster homes is considered to be therapeutic and 5 are specialised in youth with behaviour problems. Besides, studies on the incidence rate of mental disorders in children in family foster care are not available to be able to draw conclusions. Attempts to reduce the number of residential placements and boost family foster care have so far failed to achieve the expected results: In Spain, at least 28% of under 12-year-olds in residential care are awaiting foster parents and 22%, adoptive parents (López and del Valle 2015). Regarding children in care, 47% are girls and the main age groups are 12–14 year-olds and 15–17 year-olds (24% and 28%, respectively). Foreign-born children are overrepresented (22% of the in-care population compared to 15% in the general population), and most of them are in residential care (DGAIA 2016).

In Spain, the large number of children in formal kinship care also has Mediterranean cultural roots and is not centred on fostering children from ethnic minority backgrounds, as occurs in other countries. In the study by Del Valle et al. (2009), it was observed that kinship care in Spain included greater rates of family reunification and a lower rate of placement disruptions compared to non-kinship care. It offered the greatest stability of the three most common forms of foster care (kinship, non-kinship and residential care). Nonetheless, although children may have experienced more instability in non-kinship care, these placements often become long term, and in many cases even end in adoption by caregivers.

Method

Study Design

A mixed methodology was implemented including quantitative and qualitative data collection, applying complementarity as one of the possible combinations of results from different methods (Greene et al. 1989). Thus, the qualitative analysis phase was useful to clarify and exemplify the results provided by the quantitative analysis.

Participants

A cross-sectional study was conducted in Catalonia directed at all children in residential, kinship and non-kin care between 12 and 14 years of age (N = 1198) registered in the administrative database of the Catalan Child Protection System. The overall response rate was 58% (N = 700). Sixty-two cases were not included in the analysis because they had failed to answer three or more items in one, or both, of the two multi-item scales within the same psychometric scale used in the study (Personal Well-being Index-School Children (PWI-SC) and Students’ Life Satisfaction Scale (SLSS)). The cases not included did not display any significant differences in relation to the main sociodemographic variables (gender, age and nationality) compared to the sample used for the study. Accordingly, the sociodemographic characteristics of the sample were compared to the total in-care population by gender (46.1% of girls in the sample compared to 44.1% of girls in the total of in-care population); age (27.1% compared to 26.1% of 12 year-olds; 29.3% compared to 30.9% of 13 year-olds; 34.8% compared to 33.8% of 14 year-olds and 8.8% compared to 9.2% of 15 year-olds), and type of placement (59.4% compared to 56.5% in residential care; 34.3% compared to 36.1% in kinship care and 6.3% compared to 7.4% in non-kinship foster care). No significant differences were found. Multiple imputation using regression (SPSS23) was used to calculate the remaining missing values on the scales.

The sample used in this study was made up of 638 children, of whom 379 were in residential care (i.e. 69.7% of the total number of children in this age group in residential care), 219 were in kinship care (i.e. 42.8% of the total number of children in this placement type) and 40 were in non-kinship foster care (i.e. 36.6% of the total number of children in this age group in non-kinship care). In addition, 58 of these children gave a written response to an open-ended question in the last part of the questionnaire evaluating their out-of-home placement. A qualitative analysis was made with these data.

Procedure

The questionnaire was sent by post via the Catalan Government’s Observatory on Children’s Rights directly to the children (in their name) in their family and residential placements, together with a letter explaining the study. All the directors of the children’s homes and the family support teams were contacted previously to inform them about the study and data collection process, and to ensure the questionnaires could be completed by the children individually and voluntarily with informed consent and that no incentive would be given for participating.

Measures

The International Survey of Children’s Well-being (ISCWeB; www.isciweb.org), used in 15 countries with the general population of 12 year-olds, was used to provide a reference for interpreting the results. It included 34 questions organised in different subject areas related to the children’s lives, their household and people with whom they lived, their relationships with friends and people in general, their neighbourhood, their school, extra-curricular activities, and how they felt about themselves.

The original version was adapted to the specific characteristics of children in care. It was tested on adolescents from the general population by Casas and Bello (2012) and the version adapted to children in care was pilot tested by Llosada-Gistau et al. (2015). In the adapted version, no changes were made to items in the psychometric scales in their original form. The main changes consisted of including the terms “residential or family foster care, educators or foster parents” instead of “family home, mothers and fathers”, and some questions related to the type of placement and visits with birth parents.

Scales included in the questionnaire and used as indicators of subjective well-being were as follows: Overall Live Satisfaction (OLS), the Students’ Life Satisfaction Scale (SLSS) and Personal Well-being Index-School Children (PWI-SC) (Rees and Main 2015; Casas 2016). The first two were context-free, while PWI-SC was based on different life domains: satisfaction with health; the things you have; relationships with people; your school; how you spend your time; how self-confident you feel, and satisfaction with the opportunities you have in life. OLS was a single-item scale (Satisfaction with your life in general), while SLSS (my life’s going well; my life is how I want it to be; I have a good life; I have what I want in life) and PWI-SC were multi-item indicators. Cronbach’s Alpha was calculated in the two multi-item scales to measure internal consistency, which was good in both SLSS (0.87) and in PWISC7 (0.82).

The correlation between the three indicators used to measure subjective well-being was positive to high, above 0.65. The PWISC7 scale correlated in 0.673 with OLS and in 0.665 with SLSS, and the correlation between OLS and SLSS4 was 0.651.

Quantitative Data Analysis

The three psychometric scales (OLS:75.1(27.9), SLSS4:62.6(27.6) and PWISC7:80.2(16.3)) were used as dependent variables while variables related to the child protection system and satisfaction with the people the children lived with were used as independent variables. Regarding the latter, not only did they include variables obtained directly from the administrative database, but also variables in the questionnaire referring to satisfaction shown by children towards their caretakers or peers. Variables constructed from the database were operationalised as categorical variables:

  • Time spent in the last placement (less than a year; 1 < 3; 3 < 5; 5<. In addition, this variable was incorporated in the multiple regression models as a dichotomous variable (<1, 1<) because the different regression models adjusted better.

  • The number of placements experienced by the child since entering the protection system (One and two or more).

  • Previous family placement breakdowns (No breakdowns, and one or more).

  • The number of places in each residential home (<20 places; 20 < 29 places and 30<)

  • Number of times the child had run away from the residential home (0; once or more).

Variables relating to satisfaction with the people with whom children in residential or family care were living:

  • The caregivers are nice to me (disagree, agree, totally agree)

  • The caregivers listen to me (disagree, agree, totally agree)

  • Satisfaction with caregivers (not satisfied, satisfied and highly satisfied)

  • Satisfaction with peers in residential home (not satisfied, satisfied and highly satisfied).

All these satisfaction variables were measured with 11-point scales (where 0 = not satisfied or disagree and 10 = highly satisfied or totally agree). They were finally re-ordered in three categories (0–4 points, not satisfied or disagree; 5–8 points, satisfied or agree, and 9–10 points, highly satisfied or totally agree). The main reason for recoding these variables was that the categories were not evenly spaced since the distribution of these satisfaction variables did not follow a curve of statistical normality and, in its sense, it was especially important to identify adolescents who scored behind five points. Some authors argue that adult subjects who respond below 5 points have different characteristics (usually have depressive problems) and it is important to keep that in mind (Richardson et al. 2016). In addition, the non-kin foster care sample was small.

Mean and standard deviation were calculated for each independent variable according to type of placement. The Student’s t test and effect size (Cohen’s d: small d = .2: medium d = .5; large d = .8) measure for independent variables with two categories and ANOVA with a Bonferroni correction for independent variables with 3 or more categories were used to compare levels of subjective well-being. Six multiple linear regression models were constructed and are presented in this article. Three aggregate models were used in which the type of placement (residential care; kinship care and non-kinship foster care) variable was included as an independent variable. This enabled us to compare SWB with the different types of placement and the three regression models with children in residential foster care while adding the independent variables exclusive to this group. The same three models were tested among children in kinship care, but the table has not been included due to space constraints and for simplification. However, it is mentioned in the description of results. The three models tested on children in non-kinship care have not been presented as the sample was very small and statistical significance was not achieved. All the regression models were adjusted for age and gender. All the independent variables used in the bivariate analysis were included in the tested models. Finally, all variables that did not reach statistical significance in any of the models were rejected. It should also be noted that the time in foster care variable was introduced as a dichotomous independent variable (less than 1 year and 1 year or more) for a better fit in the different models.

These multiple regressions models gave heteroscedasticity. To correct this problem, multiple linear regression analysis was used with robust standard errors (Hayes and Cai 2007). SPSS 23 and Stata 14 software were used to conduct all the quantitative analyses.

Qualitative Data Analysis

Of all the adolescents who answered the questionnaire, 29.6% (198) wrote their opinion in the final Observations section. Content analysis (Bardin 2002) was carried out, classifying data in registration units by content similarity, and results processing by topic. Two main categories were identified: one that included topics related to the evaluation of the questionnaire and study, and another related to the evaluation of the out-of-home placement. Fifty-eight observations obtained in the latter category are shown in this study to complement the results of the statistical analysis. Qualitative analysis was conducted with the support of NVivo10 software.

Ethical Issues

Confidentiality and anonymity of the data were ensured according to Spanish Act 15/1999 on data confidentiality. Individual data was encoded to ensure anonymity. This study was approved by the department of the Catalonian Government responsible for the Child Protection System (DGAIA). Informants participated voluntarily and were not paid financial incentives.

Results

Six percent of youth in this study lived in non-kinship foster care, 34.3% in kinship care and 59.4% in residential care. Fifty-four percent were boys and their mean age was 13.3 years, without differences by placement type (Table 1).

Table 1 Distribution of study variables by type of placement
Full size table

SWB According to Type of Placement, Age and Gender

Adolescents in family care (kinship care and non-kinship foster care) displayed higher mean scores in the three subjective well-being indicators than those of the same age in residential care. Moreover, differences were statistically significant (p < 0.01) (Table 2).

Table 2 Mean values of the subjective well-being scales according to independent variable values by type of placement
Full size table

No differences by gender were found among adolescents in kinship care in any of the three subjective well-being indicators (p > 0.05), and the calculated effect size confirmed there were no differences by gender (between d > =.01 & d < =.21). In contrast, girls in residential care displayed lower scores (p < 0.05) confirmed by effect size (between d > =.31 & d < =.48), although girls in non-kin care displayed higher scores in SWB on the SLSS4 scale than boys. In this case, the difference also reached statistical significance (p < 0.05) and the effect size showed that the differences were medium in relation to SLSS4 (d = .67) and OLS (d = .48), but non-existent using PWISC7 (d = .07). Nonetheless, care should be taken in interpreting the results for this group given that the sample was extremely small.

Although a tendency was observed for subjective well-being to decline as children got older, differences in the bivariate analysis did not reach statistical significance among adolescents in kinship care and non-kinship care, and significant differences were only observed among those in residential care in the PWISC-7 scale.

Stability: SWB According to Time Spent in Last Placement and Type of Placement

Thirty-three percent of youth in this study had been in their last placement for more than 5 years (residential or family), while 22.7% for less than a year. By type of placement, 63.5% of adolescents in kinship care and 70.0% of non-kinship had been in foster care for more than 5 years compared to 12% of adolescents in residential care, and differences were significant (residual ± 2,58).

Adolescents in kinship care who had been less than a year or between one and 3 years in the same type of placement scored lower in subjective well-being than those who had been 4 years or more, a result observed in all three well-being indicators, reaching statistical significance in the SLSS4 scale (p < 0.05). Adolescents in residential care who had been 3 years or more in the same residential home displayed higher scores in subjective well-being. However, differences did not reach statistical significance in any of the three scales.

Stability: SWB According to Number of Changes and Type of Placement

Fifty-nine percent of adolescents in this study population had been in the same placement since entering the child protection system, while 14.9% had experienced three or more different placements. If we break down the data, 85.8% of adolescents in kinship care had always been with the same foster family, while only 47.6% of adolescents in residential care had remained in the same residential home since entering the care system, and the difference was statistically significant (residual ± 2,58). Adolescents in kinship care who had only experienced one type of placement displayed a higher level of subjective well-being in the three indicators compared to those who had experienced more than one, and differences were significant on the PWI-SC7 scale (p < 0.05). The effect size also supported these differences (d = 0.50). The same occurred with adolescents in residential care who had always been in the same care home, displaying a higher level of subjective well-being on the OLS scale, where differences reached statistical significance (p < 0.05) (d = 0.39).

Stability: SWB According to Previous Foster Care Disruptions and Type of Placement

Twelve percent of adolescents had experienced foster placement disruptions prior to their current residential or family placement. Eighteen percent of those in residential care and 15.0% of those in non-kinship care had experienced a disruption compared to only 2% in family care, with a statistically significant difference (residual ± 2.58). It was observed that adolescents in residential care who had had no disruptions in previous family placements had higher levels of subjective well-being. Differences reached significance in the three indicators (p < 0.05). The effect size showed that the differences were medium (between d > =.48 & d < =.55). No conclusions could be drawn regarding adolescents in kinship care and non-kinship foster care as only four and six cases, respectively, reported previous disruptions (Table 2).

Stability: SWB According to Number of Places in Residential Homes

Regarding adolescents in residential placements, 66.2% lived in homes with 20 places or more, and 32.2% in homes with more than 30, while 17.2% lived in homes with less than 11 places. Adolescents living in smaller-scale residential homes displayed higher subjective well-being scores on all three scales, with statistical significance on the PWI-SC7 and SLSS4 scales (p < 0.05).

Stability: SWB According to whether the Child Had Run Away from the Residential Home

It was also observed that 20.6% of adolescents in residential care had run away from the home at least once (no data was available for those in kinship care and non-kinship foster care). Those who had never run away had significantly higher levels of subjective well-being in the three indicators (p < 0.05). The calculated effect size showed that the differences were medium (between d > =.46 & d < =.53).

Relationships: SWB According to Satisfaction with Caregivers by Type of Placement

Most adolescents in family foster care (kinship and non-kinship foster care) totally agreed that their caregivers were nice to them, and were highly satisfied with them (over 75%). Slightly fewer totally agreed that their caregivers listened to them (around 67% for both types of placement). Children who totally agreed that their caregivers were nice to them and listened to them, and children who were satisfied or highly satisfied with their caregivers, clearly displayed a greater level of subjective well-being in the three indicators and differences were statistically significant (p < 0.05).

As for adolescents in residential care, 52.3% totally agreed that their educators were nice to them. This percentage was lower for adolescents who totally agreed that their educators listened to them (38.5%). It was also lower for those who claimed to be highly satisfied with their educators (38.0%). Finally, 41.4% were highly satisfied with their peers at the residential home compared to 12.2% who were not. Adolescents who agreed or totally agreed that their educators were nice to them and listened to them, as well as those who were satisfied or highly satisfied with their educators and their peers displayed higher scores of subjective well-being compared to those who did not agree or were not satisfied in the three indicators. Differences were statistically significant in all three (p < 0.05).

Aggregate Regression Models: SWB According to Type of Placement, Gender, Age, Stability and Relationships

Finally, six regression models were tested using the subjective well-being indicators as dependent variables. Hence, Table 3 shows the results for the three tested models for regression analysis of aggregate data, while Table 4 shows the results for adolescents in residential care. The number of placements variable was omitted since it was closely linked to foster placement disruption and time spent in last placement variables. In the model that only included children in residential care, the variables satisfaction with peers at residential home, size of home, and number of times the child had run away were added. Each model was adjusted for gender and age.

Table 3 Multiple regressions on the values of each of the three subjective well-being scales among children in care (aggregate sample)
Full size table
Table 4 Multiple regressions on the values of each of the three subjective well-being scales among children in residential care
Full size table

All six models were significant (p < 0.01) and displayed a high explanatory capacity for the variability in the three SWB scales used as dependent variables (R2 > 24%).

The first point to highlight about the three models for aggregate sampling is that, regardless of the rest of the variables included in the three regressions, adolescents in kinship care and non-kinship care displayed higher levels of SWB compared to adolescents in residential care. Furthermore, differences reached statistical significance (p < 0.05) in the OLS and SLSS4 models (Table 3).

Also noteworthy was the fact that 15 year-olds had lower SWB than 12 year-olds, and the difference was statistically significant in all three models (p < 0.01). This was also observed for adolescents in residential care, with a statistically significant difference in the SLSS4 model (Table 4). In the case of adolescents in kinship care, the difference in the PWISC7 model was also statistically significant.

Regarding gender, girls were seen to have lower levels of SWB than boys in the OLS aggregate sampling model (p < 0.01). Children in residential care displayed lower SWB, reaching statistical significance in all three models (p < 0.01). In contrast, girls had higher levels of SWB than boys in the PWISC7 model. No gender differences were observed in the other two models (OLS and SLSS4).

The longer the time spent in the last placement the more positive impact it seemed to have on the subjective well-being of adolescents in care, reaching statistical significance in the SLSS4 model (p < 0.05) (Table 3). And the same occurred for youth in kinship care. In contrast, differences failed to reach statistical significance among adolescents in residential care in any of the models (Table 4). Having experienced disruption in a family placement before being placed in residential care appeared to have a negative impact on subjective well-being compared to not having experienced any disruption, with a statistically significant difference in the three models, and also in the aggregate models.

Adolescents in care who were highly satisfied with their caregivers (family care) or educators (residential care) displayed higher levels of SWB compared to those who were dissatisfied. Differences reached statistical significance (p < 0.05) in all three aggregate sampling models and in the models for youth in residential care. This tendency was also the same for youth in kinship care, although statistical significance was only reached in the OLS model. Similarly, adolescents who totally agreed that their caregivers or educators were nice to them displayed higher levels of SWB, reaching statistical significance in PWISC7 and SLSS4 in both the aggregate models and in the model for youth in residential care (p < 0.05). In contrast, differences were not statistically significant in any of the three models for the kinship care sample. However, it should be pointed out that only 5 adolescents in kinship care did not agree that their caregivers were nice to them. Regarding whether youth in care felt that their caregivers or educators listened to them, greater levels of SWB were observed among those who totally agreed, reaching statistical significance in the PWISC7 and SLSS4 models of the aggregate sample (p < 0.01). Although statistical significance was not reached in the SLSS4 model for this variable among youth in residential care, the difference was significant among youth in kinship care.

Specific Factors Pertaining to Residential Care: Relationship with Other Children in the Same Placement, Number of Places, and Running Away from the Home

Adolescents living in residential care homes with fewer than 20 places were seen to have greater levels of subjective well-being than those who lived in homes with between 20 and 29 places (statistical significance observed in the SLSS4 model). They also displayed greater SWB than those living in homes with 30 places or more. In this case, statistical significance was reached in all three models (Table 4).

Finally, not having run away from the residential care home was related to higher scores of subjective well-being among adolescents in residential placements. Statistical significance was reached in the SLSS4 model (p < 0.01) (Table 4).

Factors Influencing Well-Being Based on Qualitative Data: Relationships with the People they Live with, Instability and Life Satisfaction According to Type of Placement

With regard to the qualitative analysis, 58 of the 198 adolescents who completed the open question at the end of the questionnaire referred to their personal situation and experience in the child protection system. This category was grouped into 5 sub-categories: negative evaluations regarding residential home; positive evaluations regarding their stay in residential care; positive evaluations regarding family foster care; their wish to return home, and proposals for child protection services to consider (Table 5).

Table 5 Categorisation of observations made by children regarding their personal situation and experience in the child protection system
Full size table

Links between these results and those obtained in the quantitative analysis were particularly notable in the following topics: their relationships with the people who lived with them; placement changes or instability within the protection system, and lower life satisfaction among adolescents in residential care.

Thus, seventeen negative evaluations of life in residential care were observed, some of them related to the treatment received in the home by educators or their peers that affected their well-being, such as the following example:

“I think they should keep an eye on the people they employ because some educators make us suffer a lot” (14-year-old girl, residential care)

Another aspect that expanded on the results from the quantitative analysis was stability and the wish to have the same tutor for a longer time:

“To be honest, I don’t think it’s good that we have to change tutor every year because of shift changes or whatever” (14-year-old girl, residential care)

These criticisms reflected the idea that some adolescents were not satisfied with their residential placement and some described this feeling as one of dissatisfaction with life:

“For you this is just the questionnaire of a girl in residential care, but for me it’s a sad life and I hope 1 day you can help kids like me, who spend all day crying because of their horrible life on the street and in the children’s home. Thanks” (14-year-old girl, residential care)

Also reflected was a wish for an alternative to their situation:

I’d like to return home when my dad’s personal problems have finished, or go and live with my mum. It’s not that I’m treated badly in the children’s home, but I’d rather live in my own home than in a children’s home” (13 year-old girl, residential care)

“I’d like to have another life. Go home with all my brothers and my parents and be a happy family” (12 year-old boy, residential care)

Seven positive comments were obtained regarding life in residential care, mainly related to stability and being well treated by their educators and peers:

“In general, everything’s going very well; the educators are really nice to me and I’m happy, and the boys and girls – perfect” (13 year-old boy residential care)

Adolescents in kinship care also showed appreciation for being cared for and loved and they evaluated family placements as the most positive option:

“I have a life that other children in care don’t have. I’ve got a better deal because my grandparents take care of me, and not people that I don’t know. Of course I’d like to be with my mum but if that’s not possible, I’ll stay with my grandparents who’ll continue to care for me” (13 year-old boy, kinship care)

Finally, and regardless of placement type, were a group of proposals put forward by adolescents to improve services. Apart from topics already mentioned, we can also highlight comments related to visits, having greater autonomy and influence in decision-making, or having greater financial support:

“They should give us more financial support and talk about family issues” (13 year-old girl, kinship care)

“I’d like to change some things in my life and I don’t want so many visits with my parents, and I’d like them to let me do more things on my own” (14 year-old boy, residential care)

Discussion

This study shows how some factors of the child protection system may influence the subjective well-being of adolescents in family foster care (kinship and non-kinship) and residential care. Regarding the first objective, it is worth highlighting that adolescents in foster families (kinship and non-kin) displayed higher scores in subjective well-being than those in residential care. This coincided with results from the qualitative data, which revealed more negative evaluations of life in residential care than positive, compared to only positive views of kinship care, These results are in keeping with studies which show that residential care may offer a less normal family setting (Casas 1997). They also coincide with the positive results observed for children in kinship care in some countries (Farmer and Moyers 2008; Montserrat and Casas 2007), although the issue of SWB is not dealt with in depth in these studies. Furthermore, it was observed that girls in residential care displayed lower levels of subjective well-being than boys in the same situation (also in Dinisman et al. 2012), which would appear to indicate that girls are more susceptible to situations of vulnerability (Tomyn 2013). It could also explain why significant differences by gender in well-being levels among adolescents were not found in both types of foster families (more similar to those in the general population).

Differences between residential and family foster care can be explained by factors derived from the social and cultural context in Spain. Firstly, although residential homes in Spain are not considered therapeutic placements (see González-García et al. 2017), some children in residential care naturally have mental health problems (apparently more than in family foster care, although no evidence exists for this in the Spanish context). It can be said, however, that children with psychological and behavioural problems are less likely to be placed in foster families than children without them. Moreover, children that have experienced family breakdowns usually end up in residential care. So, we are not always comparing children with the same characteristics when we compare type of placement.

It is also worth highlighting that 15 year-olds displayed lower levels of SWB than 12 year-olds. Studies with children from the general population have indicated that as adolescents get older, their level of subjective well-being decreases (Casas et al. 2013; Tomyn 2013). Thus, in Report Card 13 (UNICEF 2016), 15 year-olds reported a lower level of life satisfaction than 13 year-olds who, in turn, had lower scores than 11 year-olds. In this case, therefore, the sample would appear to reflect the behaviour of the rest of the population of the same age. However, perhaps one reason underlying decreases in SWB over time could be due to the instrument used for self-reporting, where adolescents are better at metacognition than younger children and can report with more nuances on how they feel and evaluate their lives. The use of adolescent self-report remains a controversial issue in the scientific community. Differences by age and gender were not found from the qualitative data in this study.

Explanations regarding results from the second (stability) and the third (relationships) objectives were interconnected and reflected in both data sources. From the quantitative analysis, factors regarding the positive influence stability exerts on well-being were: not having experienced disruptions in previous family placements, and having stayed longer in the same placement type. From the qualitative data, having avoided changes of placement and the wish to have the same caregiver for a longer time were the main positive factors. This is in line with a study concerning the perceptions of adults with a care background by Unrau (2010), who suggested that the move experience was important not only for the number of changes, but also for the quality of the move.

Factors concerning the relationships domain with a positive impact on well-being from the quantitative data were: being highly satisfied with their caregivers (kinship and non-kinship parents), and being treated well and listened to by their caregivers/educators, whatever the type of placement. Highlighted in the qualitative data were: being well treated by their caregivers and peers, and being cared for and loved by foster carers. Thus, stability and interpersonal relationships with key people in their lives were highly relevant factors contributing to subjective well-being at the ages studied. Low placement stability (either foster families or residential) led to similar results among all youth in care. Adolescents in foster care tended to have stayed longer in the same placement compared to those in residential care, and this may also have contributed to higher levels of SWB among them. Increasingly, greater consensus exists among authors who pinpoint stability in children’s lives as a key factor for their well-being and transition to adulthood (Biehal et al. 2015; Del Valle et al. 2009; Wade et al. 2011). In addition, this result underlines the importance of establishing a stable relationship based on trust with caregivers, in keeping with results obtained by Melendro et al. (2016), but stability is essential in order to achieve this. Bravo and Del Valle (2003) already indicated that children in residential care confided in their peers and educators at the residential home more than in those at school, and Martín (2015) also showed the importance of relationships within the context of the residential home. The qualitative data revealed that the emotional bond established with their grandparents, foster parents or caregivers when conditions for stability were in place was highly valued. It was also observed in the quantitative data that adolescents in residential care who were satisfied with their caregivers and peers displayed higher scores in subjective well-being. The high concordance between data sources highlights these results.

It was also noted that living in small-scale residential homes had a positive impact on subjective well-being, probably facilitating a more stable and personalised relationship with caregivers. Not having run away from the residential care home was also related to higher scores of subjective well-being in residential placements. However these results cannot serve to substantiate Rees (2011) when he stated that running away from the home was a way of expressing opposition to or dissatisfaction with the home.

One relevant result was that a longer period of time spent in the same out-of-home placement boosted children’s well-being, contrary to the widespread opinion among professionals and policy-makers that the longer the time spent in the protection system, the worse it was for the child. Recent studies (Biehal et al. 2015) have already suggested that if conditions in the out-of-home placement are stable and meet quality standards, it is preferable for children to remain in care than return to their birth parents, provided they are able to take care of them. The main objective of the child protection system, therefore, should be to seek the well-being of the child population, whether they are with their birth family or in an alternative out-of-home placement, always taking into account the best interests of the child as contemplated in the UN Convention on the Rights of the Child.

Limitations

This study has some limitations. First, the sample of children in foster care was too small. The issue of causality requires greater clarification. The study shows levels of SWB among children in different types of placements and with different experiences in the child protection system, but not the other way round. In other words, levels of SWB before entering the system cannot be known, nor if the child’s SWB actually determined the factors related to stability and relationships. Another issue was the area of study. Despite the situation in the child protection system being similar throughout the whole of Spain, data collection was only carried out in Catalonia.

Implications

Nonetheless, this study opens the door to debate and puts forward some recommendations to be taken into account in policy-making and professional practice. This study has reaffirmed the idea that, in terms of SWB, it is better to be brought up in a family setting (be it kinship or non-kin care) than in residential care. However, the type of placement in itself is not sufficient, but will depend on conditions of stability and on trust relationships. This means, firstly, avoiding changes of placement for administrative or political reasons, and working more on a child-centred basis, ensuring that caregivers and foster families are trained for the role and committed to care for the child. But it also entails the need to improve visits with siblings and parents, the importance of giving children greater autonomy, avoiding overprotection when they are in care, and allowing them to take part in the decision-making process. Child protection systems should offer quality care and adopt a cross-cutting approach in which these factors are clearly taken into account throughout the entire service network.