1 Introduction

1.1 Children in Public Care

According to Catalan law (LDOIA 14/2010), children in public care are those who at some point in their life are in a situation where they lack the basic elements for their integral development within the family unit and need to be protected by the competent authority and placed in foster family or residential care. In most cases, entry into the protection system is the result of suffering some form of abuse within the family. The most prevalent types of abuse in Catalonia (Departament de Benestar Social i Família 2014), as in most Western countries (Gilbert et al. 2009) are, in order of most to least number of cases detected and reported: neglect, psychological abuse, physical abuse and sexual abuse.

The same law states that if a situation of helplessness is detected and it is necessary to separate the child from their family, the type of care must take into account the best interests of the child and therefore preferably grant care to extended family, followed by a foster family with no relatives and finally residential care. In 2014, 57 % of children were in family foster care (of which 71 % were kinship care and 29 % non-kinship foster care), while the remaining 43 % were in residential centres (Departament de Benestar Social i Família 2013). This large number of children still in residential centres is a feature and a handicap of child protection systems in Spain. Del Valle et al. (2009) have already pointed to the limited development of non-kinship care programmes. López and Del Valle (2015) found that children under 12 who were in residential care were mainly groups of siblings or children with behavioural problems or disabilities, for whom it was not possible to find foster families. The picture is very different in the rest of Europe, where on the one hand, in the Nordic countries and the UK residential care seems to be reserved for adolescents with behavioural problems (Jackson and Cameron 2014), and at the other extreme we find Portugal, with most cared-for children in residential centres (Canali and Vecciato 2013).

We present the following data in order to contextualize our study. Focusing on the characteristics of children in care aged 11 to 16 in Catalonia (Montserrat et al. 2015), 55 % are boys and 45 % girls, and 17 % were born outside Spain. An increase can be observed in the number of children in the child protection system as age increases within this range, from an average of 16 % in the 11–12 year age range up to a quarter of those in the 15–16 year age range. Half of these children have been in the care system for four years or more and 40 % have been in two or more types of placement.

We must also take into account other characteristics of children in residential centres already identified in other studies, such as those conducted by Fernández-Daza and Fernández-Parra (2013), which show that, in general, these children have fewer psychosocial skills than the general population of the same age, more severe behavioural problems and more socio-emotional alterations. Other studies have found that children in residential care display delays in their development compared with the general population (Carbone et al. 2007) or important social and psychological delays (Chaves et al. 2013).

Another key aspect concerning the in-care population is the evidence that their pathway through the education system is generally unsatisfactory (academic delay, school dropout, etc.) and that their family background and lack of prioritization of academic results by the care system influence failure at school, exposing them to the risk of social exclusion (Montserrat et al. 2013a, b; Jackson and Cameron 2014; Fernández-Daza and Fernández-Parra 2013). For example, in Catalonia (Montserrat et al. 2013a, b), only 31.1 % of 16-year-old adolescents in residential care are in the school year that corresponds to their age, compared to 37.4 % of those in non-kinship foster care, 42.1 % in extended family care and 73.5 % of the general population. At the age of 12, the same rate for those in care stands at 57.4 % compared to 92.7 % of the general population, and the rate of adolescents in care in special education schools at 15 years of age stands at 12.2 % compared to 1.1 % of the general population. In a study on the social integration of children in residential care in the context of school, Martin et al. (2008) found that children in residential centres suffered a higher rate of rejection by classmates than other students when being chosen for academic tasks, although they found no significant difference in the rejection rate when it came to leisure activities. Classmates of children in centres attributed characteristics to them that explained this rejection: having bad relationships with teachers, aggressiveness, attention-seeking and not being noted for their ability to resolve conflicts or their maturity.

However, in Sweden Henin (2013) found that children with real academic difficulties and school absenteeism prior to non-kinship foster care experienced a positive change towards school after being in care in two main areas: socialization (having a good relationship with classmates fostered self-esteem) and learning (high performance fostered their being both happier and prouder). The author pointed out that the carers played an important role in these changes towards education: routines like sleeping enough, regular meals, support and the foster family liaising with teachers.

Another key area to take into account is the one highlighted by Fernandez-Daza and Fernandez-Parra (2013), who also found significant differences between children in care and their peers of the same age in the general population in terms of the number of times per week they shared time with, played with or did activities with friends.

A final key aspect to point out is the increasing consensus regarding stability being one of the key aspects that favour the well-being of children in care: that is, when they can keep the same friends, go to the same school and do the same extracurricular activities, among other things (Wade et al. 2011; Biehal et al. 2015; Montserrat et al. 2013a, b). Children in kinship care experience a more stable form of foster care and can reach adulthood accompanied by their family, two aspects that probably make their pathways more similar to those followed by the general population (Montserrat 2012; Del Valle et al. 2009). In contrast, those in residential care often perceive an uncertain future, which provokes feelings of insecurity, together with a strong sense of temporariness (López and Del Valle 2015).

1.2 The Children’s Subjective Well-Being

Subjective well-being can be considered as an essential component of the broader concept quality of life. Conceptually, quality of life is composed of material and non-material aspects. Children’s subjective well-being is a non-material, psychological component of quality of life: children’s perceptions, evaluations and aspirations on their lives, which we usually assess considering life domains such as health, leisure, school, interpersonal relations, personal satisfaction, and so on (Campbell et al. 1976; Casas 2011; Diener 2012).

Studies that have measured the level of SWB in different countries have found that scores among 12-year-old children are higher than those of at more advanced ages (Casas et al. 2012; Casas et al. 2013; Tomyn 2013; Viñas et al. 2015). Moreover, it has generally been observed that at these ages girls score lower on SWB than boys. This lower SWB for girls has usually been attributed to psychosocial factors, such as, for example, a greater presence of depression among girls, less self-esteem or less satisfaction with their physical image (Grabe et al. 2007; Frydenberg and Lewis 2009; Viñas et al. 2015).

Rees et al. (2011) also explored the SWB of vulnerable children, studying the relationship between the SWB of children aged 8 to 15 and the socioeconomic status of their families in the English context. They found a lower SWB among the more disadvantaged classes, those whose parents did not work and those with families with lower incomes. The authors concluded that children’s SWB is associated with the family’s economic situation and that children are sensitive to changes in the family economy and issues concerning their parents.

In addition, Tomyn (2013) found that children in Australia who had experienced domestic violence and unstable living arrangements had a lower SWB than the general population. “At risk” adolescents were twice more likely than the general population to suffer or have a high risk of suffering from depression and scored lower than their peers at the same age, especially in areas such as standard of living, future security and health. Girls “at risk” scored lower on all of the items studied.

1.3 The Subjective Well-Being of Children in Public Care

Returning to the central topic of the study, the first thing to note is that of those studies that focus their interest on children in public care, few emphasize the importance of their subjective well-being (SWB). And from the perspective of research into subjective well-being, very few studies are aimed at children in care. In this section, we identify some of those that do.

Studies by Llosada-Gistau et al. (2015) in Spain and Schütz et al. (2015) in the Brazilian context found that children in residential care obtained lower mean SWB scores than the general population of the same age. The results of the former highlighted the need to improve satisfaction with particular life domains, such as school and residential homes, particularly among girls in residential centres. The latter authors found that satisfaction with family and home were the items that contributed most to differentiating children in residential care from those who lived with their families. They also found that more time spent in residential care did not lead to a lower SWB. In the same vein, studying a sample of Spanish children in residential care, Dinisman et al. (2012) observed a tendency for them to have less stability than boys and girls who lived with their families, and that the changes experienced by the children in care adversely affected their SWB, especially their interpersonal relationships, academic achievements and leisure time.

As for those in kinship care, Montserrat and Casas (2007) showed that children fostered in this type of placement were satisfied with their lives and obtained similar results to those of the general population in five life domains. While in kinship care, children are generally allowed to maintain contact with the rest of their family and keep in touch with their roots, which allows them to consolidate a sense of belonging and identity (see also in studies not focused in SWB like Farmer and Moyers 2008; Burgess et al. 2010).

Our aim here is to provide a more detailed study of SWB among children in care, given the gap in the reviewed literature, and analyse type of placement in greater depth, as we have not found studies that explore the differences between residential and foster care with regard to SWB. Thus, the aim of our study is to determine the subjective well-being of 12 to 14-year-old children in foster care - extended family, non-kinship or residential - in Catalonia, assessed using the Personal Well-Being Index-School Children, and to analyse how it correlates with several explanatory variables.

2 Method

2.1 Sample and Instruments

We carried out a cross-sectional study in Catalonia in 2013. The study population comprised children in care aged between 12 and 15. The data were obtained by administering the International Survey of Children’s Well-being questionnaire (ISCWeB; www.isciweb.org), which had been adapted to fit the unique characteristics of children in care (Llosada-Gistau et al. 2015). We used responses to 34 questions grouped into 12 thematic sections: personal information, home and people you live with, money and possessions, friends and other people, the area where you live, school, free time, your life, how you feel about yourself, more about you, your home, and the questionnaire itself. The Catalan version of the ISCWeB had already been tested by Casas, Bello et al. (2012). A few changes were made to adapt the questionnaire to the in-care population, and this version was tested in a pilot (Llosada-Gistau et al. 2015). No change was made for the psychometric scales. Modifications for the in-care population were mainly made in the section on “Where do you live and your home”, where “family” was replaced with “residential care or family foster care” and “parents” were replaced with “carers”; four questions regarding the type of placement they were in were also added. In 2013, the questionnaire was sent to all children in care in Catalonia born between 1998 and 2000 (N = 1198) and registered in the regional government’s database; 58 % responded (700). We therefore used the list of all children in the selected age-range in residential, kinship or family foster care in Catalonia. The aim was to contact all of them and ask whether they would agree to participate in the research. However, access to children in non-kinship foster care was much more difficult than for those in residential care or kinship care due to some of the gatekeepers (professional teams) not forwarding the questionnaire to them. Our sample is therefore ad-hoc and includes all of those children who responded. We compared the sociodemographic characteristics of the sample regarding gender, place of birth and education with the overall population in care and found no significant differences.

From the initial sample of 700 children, 31 cases were not included in this analysis because 3 or more items on the Personal Well-Being Index-School (PWI-SC6) were missing. These cases did not show any relevant distinct characteristics compared to the other children. The remaining missing data from the scale items were imputed by regression. The final in-care population sample comprised 669 children, distributed as follows: 397 in residential care (73 % of all those in centres), 251 in kinship care (49 % of the total for this type of care) and 41 in non-kinship foster care (37 % of the total for this type of care).

The ISCWeB questionnaire is self-administered and includes a modified version of the PWI-SC. The PWI-SC scale was designed by Cummins and Lau (2005) to measure SWB among children and is based on the original PWI (Cummins et al. 2003), which measures subjective well-being among adults. The PWI-SC originally included seven items on satisfaction with different life domains.. Satisfaction with the things you want to be good at and Satisfaction with doing things away from your home were not included because when piloting the questionnaires with children in care these items displayed poor understanding. In this study, an additional item was included on satisfaction with use of time, as proposed by Casas et al. (2012). We shall call this version the PWI-SC6. Another item on satisfaction with school, proposed by Tomyn and Cummins (Tomyn 2013), which is sometimes added to this scale, has not been included in order to avoid redundancy in the planned regression analyses using independent variables related to school satisfaction. Therefore, PWI-SC6 includes satisfaction with the following items:

  • your health

  • how secure you feel

  • the opportunities you have in life

  • the things you have

  • your relationships in general

  • your use of time

The score for each item ranges from 0 to 10 points and only the end values are labelled. This version, validated on children from the general population by Casas, Bello et al. (2012), was validated on children in residential care by Llosada-Gistau et al. (2015), showing good psychometric properties. A review of the scientific literature, as we have seen in the introduction, reveals that few studies have explored the satisfaction of children in these situations, much less used scales previously validated on the general and in-care populations, which is why we chose this version of the PWI-SC.

The PWI-SC6 scores are used as indicators of SWB in this study. Scores are the result of summing up the six items, linearly transformed to a 0 to 100 scale.

The independent variables included in the study were selected according to two main criteria. On the one hand, those related to personal characteristics and variables that reflect children’s perceptions and evaluations of different aspects of their life. These are also used in SWB studies on children from the general population and address key domains in children’s lives such as school, leisure time, friends and critical changes (Casas 2011; Casas et al. 2012; Dinisman et al. 2012). The following variables were included: gender, age, born in Spain, education, change of school or carer in the last year, liking going to school, feeling safe at school, frequency of extracurricular activities, frequency of going out with friends, frequency of being left out by other children in their class in the last month, having access to a computer, and satisfaction with classmates and marks. On the other hand, we included variables related to these children’s situation in care that may be important in understanding and analysing their well-being: type of placement, happy with their type of placement and frequency of contact with mother and father.

2.2 Data Collection

The questionnaire was sent by ordinary mail form the Welfare Department to children in residential and kinship care in their own name. The envelope included a letter with a brief description of the study and the questionnaire. A contact email was also added to answer any questions regarding the questionnaire. In order to encourage participation, we also contacted the directors of each residential centre to inform them of the process and the purpose of the study. To protect the confidentiality of the answers, we asked the directors to ensure the appropriate precautions to keep the survey answers confidential (see ethical considerations below).

As for the children in non-kinship care, we contacted the teams of professionals dealing with their case to inform them of the study. These teams contacted the foster parents to explain the study and sent the questionnaire to the children, ensuring that they responded to the questionnaire autonomously and confidentially.

As it was sent to all children in care of the age groups mentioned, those with special developmental or behavioural concerns were also included. These children, who are mostly concentrated in residential centres, received help answering the questionnaire, including reading the questions and writing if needed. The idea was not to exclude anyone for any reason. Although those questionnaires answered with outside help are identified, the data analysis could not contemplate the variable of whether they have special developmental or behavioural concerns due to this not having been encoded. Neither is there a variable identifying this situation in the regional government’s database.

2.3 Data Analysis

In all the analyses presented here, the PWI-SC6 is considered the dependent variable and the rest independent. The mean and standard error of the PWI-SC6 scale were calculated for each independent variable according to type of placement. The Student’s t-test, ANOVA, Chi-square test and linear regression were used. We have constructed two multiple linear regression models, with Type of placement as independent variable in one model and as a stratifying variable (aggregate group, kinship care group and residential care group) in the other. In the first model, we aimed to determine how the SWB of children in care varies with Type of placement, regardless of other independent variables. In the second the aim was to see whether the independent variables affect SWB unevenly according to Type of placement.

In the multivariate analysis, we decided excluding non-kinship foster care because the sample for those in non-kinship foster care is very small.

Although we included all independent variables in both models, the tables only reflect those variables that proved significant in the multivariate model. However, we adjusted both models for gender and age. Both models showed heteroscedasticity, which we compensated for using robust multiple linear regression.

2.4 Ethical Considerations

Data confidentiality was guaranteed and individual data were encoded to ensure anonymity. This study was approved by the relevant department of the Catalan government (DGAIA), which is the body within the public welfare system responsible for consent with regard to the participation of children in care (residential centres and foster families) and ensuring the issue of privacy and confidentiality. The survey was sent by the Welfare Department to the addresses registered on their database. Children participated in the study voluntarily, only under consent and without any kind of incentive.

3 Results

Of the 669 children in the sample, 54.3 % were boys (53.9 % residential care; 52.4 % kinship care and 68.3 % non-kinship foster care). By age, 26.8 % were 12 years old, 29.9 % 13 years old, 34.2 % 14 years old and 9.1 % 15 years old. By Type of placement, the distribution of ages is similar (mean 13.3 years old). 84.6 % were born in Spain, unevenly distributed: 21.7 % of those in residential care were foreign compared to 5.7 % of those in kinship care (Table 1).

Table 1 Adolescents’ personal characteristics and perceptions and evaluations of different aspects of their life according to type of placement
Full size table

Children’ perceptions and evaluations of different aspects of their life:

  • Satisfaction with type of placement. 14.9 % of the children in care in our study were not very or not at all happy with the type of placement, with significant differences depending on whether they were in centres (23.7 %) or kinship care (0.9 %) (P. residual +/−2.58). 65.3 % were quite or very happy, also with significant differences, those in kinship care (95.1 %) being more satisfied than those in centres (46.2 %) (P. residual +/−2,58) (Table 1).

  • Contact with parents 69.1 % of the children had contact with their mother, more frequently those in centres (78.2 %) than those in kinship (55.1 %) and non-kinship foster care (60.5 %); 17.1 % had no contact, while 13.9 % did not have a mother (due to death, disappearance, etc.). Among the latter, differences are significant (P. residual +/−2.58) according to whether they are in centres (7.8 %), extended family (21.2 %) or non-kinship family (31.6 %). 50.2 % of the children in care had contact with their father, 26.3 % had no contact while 23.5 % did not have a father. Significant differences are observed in the fact of not having a father between those in kinship and residential care (32.6 % and 17.5 %, respectively) (Table 1).

  • Changes of family and school during the last year and being in school year corresponding to age. 69.0 % of the children had maintained the same family or carer in the previous year. 48.2 % of those in residential care had changed carer compared to 6.1 % of those in kinship care and 7.3 % in non-kinship foster care (P. residual +/− 2.58). Although many students change school at 12 to go from primary to secondary school, there are large differences in the number of changes, depending on whether they are in centres (57.0 %) or extended family (22.1 %) (Table 1).49.5 % of the children in care were in the academic year that corresponded to their age. 50.8 % of those in residential care had repeated at least one year compared to 39.9 % of children in kinship care and 43.9 % in non-kinship foster care. Most children in special education are in residential care (6.1 % compared to 0.9 % in extended family) (Table 1).

  • School. 60.4 % of the children in care agree or strongly agree that they like going to school compared to 17.6 % who disagree or strongly disagree. 83.0 % agree or strongly agree that they feel safe at school and in neither of the two items were differences found according to Type of placement. Adolescents in care score an average 7.8 points in satisfaction with their classmates, with significant differences (p < 0.01) according to whether they are in centres (7.6), extended family (8.2) or non-kinship care (8.4). As for satisfaction with marks, the mean score was 5.8 points, with no differences according to Type of placement (Table 1).

  • Free time activities. 56.2 % of the children in care did extracurricular activities at least once a week, which rises to 73.2 % for children in non-kinship foster care. 47.8 % of the children in care went out with friends not often or never (57.8 % of those in residential care and 33.2 % of those in kinship care, these differences being significant). 25 % of the children in care had been left out one or more times in the previous month, with no differences according to Type of placement. 45.9 % of those in residential care used the computer every day or almost compared to 57.2 % and 63.4 % of those in kinship and non-kinship foster care, respectively (Table 1).

3.1 The Subjective Well-Being of Children According to Independent Variables

  • SWB according to type of placement. The mean score on the PWI-SC6 scale stands at 80.70 out of 100 for children in care. Those in residential care have a lower SWB (75.87) than those in kinship (87.91) and non-kinship foster care (86.78), with significant differences (p < 0.01).

  • SWB by gender, age and place of birth. Boys display a greater SWB (81.66) than girls (78.25), with significant differences (p <0.01), especially for those in centres (p < 0.01). Although a decreasing-with-age trend in the SWB scores is observed, differences do not reach statistical significance in any of the two regression models: 12-years old children scored 83.37 points compared to 79.24 for 14 year-olds. No significant differences were observed for having been born in Spain (Table 2).

    Table 2 The Subjective Well-Being of adolescents (PWI-SC6) according to independent variables (mean and standard error). Results of ANOVA, T-Tests and Linear regression
    Full size table

  • SWB by satisfaction with type of placement. Those children who were quite or very happy with their type of placement scored a mean of 86.96 points compared to 65.56 points among those who were not very or not at all happy (p < 0.01). The same effect is observed in all types of placement (p < 0.01). Also, practically all of those in family care were quite or very happy with the placement (except 11 cases in kinship care and 2 in non-kinship foster care) (Table 2).

  • SWB by contact with parents. Those children in care who had some kind of contact with their mother had a lower SWB than those who did not have a mother (due to death or disappearance) (79.81 and 85.14 points, respectively) and differences are significant (p < 0.05). No differences are observed in SWB between those having contact with their mother and those not having any contact with their mother. Having or not having contact with their father is not reflected in the different scores on the PWI-SC6 scale or the grouped sample, nor according to type of placement (Table 2).

  • SWB by changes in the family and school in the previous year and being in school year corresponding to age. Those children in care who had had the same carers in the previous year had a greater SWB than those who had changed (p < 0.01). Children in care who had changed school in the previous year had a lower SWB compared to those who had stayed in the same school (78.20 and 82.61 points, respectively) (p < 0.01). By type of placement, differences do not reach statistical significance, although regarding effect size, a small effect (−0.32) is observed among children in kinship care. Children in the school year that corresponded to their age displayed significantly higher SWB scores than those had repeated a year or years (82.87 and 79.13, respectively) (p < 0.01) (Table 2).

  • SWB by school variables. Children in care who quite strongly or strongly agreed that they liked going to school displayed significantly higher scores in SWB than those who didn’t agree much or at all (84.03 and 70.33 points, respectively) (p < 0.01). We observe the same tendency among both those in kinship care and in residential care, and in both cases, differences are statistically significant (p < 0.01). Children in care who quite strongly or strongly agreed that they felt safe at school displayed a higher SWB than those who didn’t agree much or at all (82.99 and 65.83, respectively) (p < 0.01). The greater the satisfaction with classmates, the higher the SWB scores observed, both among the grouped sample and according to type of placement; in all cases the relationship is statistically significant (p < 0.01). Equally, the greater the satisfaction with marks, the higher the SWB scores observed, both among the grouped sample and among children in residential and kinship care (p < 0.01) (Table 2).

  • SWB by free time activities. Doing extracurricular activities at least once a week increases SWB compared to doing them not often or never (85.67 and 78.17, respectively) (p < 0.01). The same effect is observed among children in kinship care and residential care (p < 0.01) (Table 2). If we calculate effect size, we observe that there is also a small effect among children in non-kinship foster care (−0.428).

Among children in care, going out with friends every day or almost or once or twice a week increases SWB compared to going out not often or never (86.44, 83.01 and 77.13 points, respectively) (p < 0.01). The same tendency is observed among children in residential care (p < 0.01) (Table 2).

Being left out by friends one or more times has a negative effect on the SWB of children in care compared to never having been left out (76.29 and 82.25 points, respectively) (p < 0.01). The same effect is observed among those in kinship care and residential care and the differences are statistically significant (p < 0.01) (Table 2).

Children in care who used the computer every day or almost had significantly higher SWB scores than those who used one few days a week or less than once a week (83.66, 80.25 and 75.89, respectively) (p < 0.01), particularly those in residential care (p < 0.01) (Table 2).

4 Results of the Multiple Linear Regression

We tested a model with the entire aggregate sample (Table 3), including type of placement as an independent variable. We excluded children in non-kinship foster care from the regression models because the sample was too small. The model is significant (p < 0.001) with an explanatory power of 48 % (R2 = 0.484). Neither gender nor age reaches statistical significance (p > 0.05). We observe that type of placement reaches statistical significance and being in residential care reduces the SWB of children in care by 3.4 points on average on the PWI-SC6 scale, regardless of gender, age and other variables included in the model compared to being in kinship care. Being quite or very happy with the type of placement increases the SWB of children in care by almost 15 points compared to being not very or not at all happy (p < 0.01). Doing extracurricular activities every day or almost relates to a higher SWB than doing extracurricular activities not often or never (p < 0.01). As satisfaction with classmates and marks increases, so does SWB (p <0.01). Finally, the fact of going out with friends every day or almost as well as using a computer every day or almost compared to not often or never also seems to favour the SWB of children in care (p < 0.01).

Table 3 Multiple regression model including type of placement as independent variable. (dependent variable: PWI-SC6 scale)
Full size table

In the second model (Table 4), we have calculated three regressions stratified according to type of placement: one for all children in care (aggregate model), one for kinship care, and one for those in residential care.

Table 4 Multiple regression models (stratified according to type of placement) (dependent variable: PWI-SC6 scale)
Full size table

Comparing regressions among children in kinship care and residential care (Table 4), we observe that both reach statistical significance (p < 0.001). However, the explanatory power of the model with children in kinship care is lower than that of children in residential care (R2 = 0.318 and R2 = 0.469, respectively). Gender and age do not reach statistical significance in either of the two regressions (p > 0.05). Being quite or very happy with the type of placement compared to not being favours SWB, among both children in residential care and kinship care. Among those in kinship care, being quite or very happy with the type of placement favours their SWB compared to being not very or not at all happy among children in residential care (13.1 and 32.6 points on average, respectively) (p < 0.01). We observe differences between the two groups in the variables doing extracurricular activities, going out with friends and frequency of computer use. None of the three variables reaches statistical significance among those in family kinship care. They are, however, variables that are significantly associated with the PWI-SC6 scale among children in residential care (p < 0.05). In the aggregate model we observe that the independent variables behave similarly to the first model.

5 Discussion

In our study, adolescents living in kinship and non-kinship foster care reported higher SWB in all life domains than those in residential care. The following may all have an influence on the SWB of children in care: gender, age, agreeing with their placement, type of relationship with their mothers, stability at school and in their placement, satisfaction with school (such as going to school, feeling safe at school, having good relationships with classmates, being in the year corresponding to their age or getting better marks) and with relationships with friends, leisure time activities and frequency of computer use. These findings represent an important step towards understanding the key factors that should be addressed in order to improve the lives of children in care and consequently issues related to the public care system.

The mean score on the PWI-SC6 scale for children in care is 80.7 points out of 100. However, the results show a much lower subjective well-being among children in residential centres (75.87) (as also observed in Llosada-Gistau et al. (2015), than those in non- kinship (86.78) and extended family care (87.91). The means for those in family foster care (extended and non-kinship) are very close to that of children in the general population. Montserrat and Casas (2007) also observed that those in kinship care obtain similar results to the general population in different life domains.

However, when drawing conclusions, we must be careful not to attribute all the responsibility for these results to the residential centres alone, because the characteristics of those in residential care differ from those cared for in foster families (extended and non-kinship): children born in other countries are over-represented in centres (22 %) compared to those in kinship care (6 %). This may be due to their not having an effective family support network or having no relatives in the country. We also observe that children who receive special education are mainly cared for in residential centres, due to the difficulty of finding them a foster family, also found in López and Del Valle (2015), where this was identified as one of the reasons for children waiting to being placed in a foster family. Canali et al. (Canali and Vecciato 2013) also observed a higher proportion of foreign children and those with mental disabilities in centres in Italy.

Some studies find a lower SWB among girls in the general population (Grabe et al. 2007; Frydenberg and Lewis 2009; Viñas et al. 2015); Tomyn (2013) also noted a lower SWB among girls socially at risk, and in our study this result is repeated for girls in residential centres, a phenomenon that requires more in-depth analysis and is important due to its practical implications. The decreasing-with-age trend, pointed out by some authors for children in the general population (Casas et al. 2012; Casas et al. 2013; Tomyn 2013; Viñas et al. 2015), is also observed in our sample, but it does not reach statistical significance.

One revealing result is that half of the children in residential care are not happy living in a residential centre and this adversely affects their SWB. This is not a minor aspect, as it refers to where they live: 95 % of children are happy to be fostered in extended family, whereas only 46 % of those in residential care are. The differences are obvious and this situation needs to be redressed, both by promoting family foster care (extended and non-kinship) and making these centres smaller and more normalized. Currently, 57 % of the centres where these children live have between 20 and 50 children, which means living in a place where less personalized care is given, differing from that received by children from the general population and with many changes in educators’ shifts. It will be necessary in subsequent research to study how the size of the centre can affect these children’ SWB.

In the first regression model, where we included type of placement as an independent variable, we found that living in kinship care increases SWB compared to living in a centre, regardless of whether children are happy with the placement. However, the fact of being happy with the type of placement has a greater impact on SWB than type of placement. Therefore, the placement is important but their perception of it is even more so, which reminds us that we need to take the opinions and perceptions of children themselves more seriously.

Another significant result stems from the bivariate analysis, which allowed us to observe that contact with mothers affects SWB: children who do not know their mother or whose mother has died obtain a higher mean score on the PWI-SC6 scale than those who have a mother, whether they have contact with her or not. One explanation for this result is that their situation is clearer and there is no need for them to form any expectations; result that can be linked to the ambiguous loss theory (Boss 1999). The same is not true of fathers. The difference between mothers and fathers contacts consequence should be explored more deeply.

In line with previous research, we confirmed here that stability, in terms of staying at the same school and retaining the same carer, increases the SWB of children in care (Dinisman et al. 2012). Many authors highlight the issue of stability as key (Wade et al. 2011; Biehal et al. 2015). In our study, children in residential care had changed carer and school in the previous year more often than children in family foster care (extended and non-kinship).

A result that awards central importance to school is the following: children in care who like going to school, feel safe at school, have good relationships with classmates, are in the year corresponding to their age or who get better marks all display significantly higher SWB scores, regardless of the type of placement. In light of these results, the child protection system should commit decisively to a policy to promote and facilitate the schooling of children in care as far as possible, in the knowledge that even among those in residential centres this acts as a highly compensatory factor. School integration favours social inclusion processes in the medium and long term (Jackson and Cameron 2014; Montserrat 2012).

Regarding free-time activities, children in care who do extracurricular activities display higher SWB scores than those who do not. This variable reached statistical significance in the bivariate analysis among children in both residential and extended family care. Therefore, given that 45 % of children in centres and kinship care do not do extracurricular activities, decisive policies should be introduced to provide incentive to do them.

In the second model, calculated according to type of placement, we found that the regression using the sample of children in residential care had greater explanatory capacity of variability for the PWI-SC scale than using the sample in kinship care (46.9 % and 31.8 %, respectively), suggesting that the variables included have a greater impact on these children’ SWB than on that of those in kinship care. In this regard, we found that frequency of going out with friends and frequency of computer use are two variables that correlate significantly in the sample of those in residential care, whereas this is not the case for children in kinship care. Presumably, there is no association between these two variables in the SWB of children in the latter type of care because both activities are much more common in this type of placement. In contrast, they are two very important activities for children in residential care and being able to do them regularly or not affects their SWB.

One of the limitations to consider in this study is interpretation of the data received from those in non-kinship care, due to it being a smaller sample, meaning their results could be biased. Precisely for this reason, in both regression models we worked without children in non-kinship foster care.

In addition, future research should also study how variables related to the protection system not included in this study can affect the SWB of children in care, such as the number of types of placement an adolescent has had, the amount of time spent in the last type of placement, the fact of having suffered a failed foster placement, etc. This study is also relevant since it supports future researches such as using qualitative methods, taking into account key aspects over their lives and developing it in other countries.

Many future challenges emerge from the results of this research, which supports proposals being made from the perspective of children’s subjective well-being. Our findings highlight: (a) the need to address the participation of children in any decision that affects their lives, as we see the importance of agreement with placement type, for example the involvement of children in the decision making process; (b) the need for a review of residential care and the material and relational living conditions of children cared for in these facilities, it is about the size of centres and the stability of the staff; (c) the need to continue promoting family foster care (extended and non-kinship) and pay greater attention to the positive results of kinship care; (d) that effort is required to clarify the role of biological parents as far as possible, as we see an unexpected result regarding the relationship with the mother; (e) that school must again be made a focal point of the lives of children in care, as we see how it can act as a compensating factor, prioritizing the education of children in care; (f) the same is true of the role of friends and free-time activities: everyday activities such as going out with friends and being able to use the computer should be promoted as standard behaviours; and finally, (g) that life stability should be guaranteed for these children whenever possible, avoiding changes of school and carer.