Legitimate child protection interventions and the dimension of confidence: A comparative analysis of populations views in six European countries

Abstract

The legitimacy of welfare state institutions is a key question in public policy research. In this study we examine population’s confidence in child protection systems, the role of institutional context and moral alignment. Analysing representative samples of survey data (N=6,043) of citizens in six European countries (Czechia, England, Finland, Norway, Poland and Romania), we find that overall people express confidence in their child protection system. Differences between populations are correlated with institutional context, i.e. the type of child protection system in place – that is, if people live in a country with a risk-oriented system or a family service-oriented system. People’s view on their moral alignment with the system (or not) only shows minor differences in support of interventions. However, a tendency towards polarisation is detected in Finland and Norway with clear differences in support of interventions that restrict parental rights: individuals who state they are in alignment with the system favour stronger interventions than those who say they are not.

Introduction

Having legitimate state institutions is an essential virtue of democratic governance, stability and regime viability (Lipset, 1959), as well as sustainability of the welfare state institutions (Roberts, 2011) such as the child protection system (CPS) (Skivenes and Benbenishty, 2022a) and social services (McCulloch and Webb, 2020). In political theory, legitimacy is often closely associated with concepts of trust and confidence (Alexiou and Wiggins, 2018; Jackson and Gau, 2015). In this paper our aim is to address two dimensions of legitimacy that are highly relevant for CPS: firstly, an institution’s claim to rightful authority (i.e. when citizens believe that the institution has a positive right to dictate appropriate behaviour and when citizens feel that they have a corresponding duty to obey, even when they disagree); secondly, as the rightfully held and exercised power (i.e. when citizens believe that the institution acts according to the appropriate norms and moral conduct; Jackson and Kuha, 2015; Tyler, 2006). The empirical analysis examines public confidence in the child protection system in six European countries: Czechia, England, Finland, Norway, Poland and Romania. We chose these countries because they differ in terms of CPS, welfare state models and rankings on child well-being measures. It is especially of interest to increase our knowledge about populations in eastern European countries as governments have been vocal critics of child protection interventions (Datta, 2018; Falch-Eriksen and Skivenes, 2019).

The obligation of states to protect children from abuse and neglect is regulated in Article 19 of the United Nations Convention on the Rights of the Child (CRC), a convention which all states, except the USA, have ratified. Article 19 of the CRC obligates countries to have a CPS that may intervene in the family if parents are unwilling or unable to care for their child. It is a system that has the authority to restrict individual freedom and make consequential and intrusive interventions into people’s family life, with a great deal of uncertainty in the outcome of the decisions made (Berrick et al., in press; Gilbert et al., 2011). There is a huge knowledge gap in this area of welfare state research (Berger and Slack, 2020; Skivenes, 2021; Skivenes and Benbenishty, 2022a) as well as in social policy discussions (Busemeyer and Lober, 2020; Pfau-Effinger, 2005; Svallfors, 2012; Valarino et al., 2018).

We address some of these gaps by examining populations’ confidence in the CPS system, and people’s meaning of moral alignment or non-alignment with the CPS. We use the theory of institutional context to shed light on findings and explore the role of sociodemographic variables. Our data material consists of representative samples of citizens in six countries, with a total 6,043 respondents.

The paper contributes to the academic debates about public policy and legitimacy in a political order by examining dimensions of confidence in welfare states. It contributes to the social policy literature on the role of institutional context and people’s perceptions on welfare state policies. The study expands the understanding of the relationship between a person’s beliefs on moral alignment with the existing CPS and their recommendation for state intervention. Finally, it contributes to the empirical, comparative knowledge base on citizens’ perception and understanding of the CPS and the government’s responsibility for children.

The paper is organized in six parts. In the next section, the theoretical platform is presented, followed by an outline of models of CPS and the empirical field. We then describe our data and method of analysis and present the findings with discussion and concluding remarks.

Conceptualising legitimacy and confidence

A challenging part of any study of legitimacy is the conceptualisation and operationalisation of the term. There is scant agreement on how to understand legitimacy among scholars of different disciplines (Alexiou and Wiggins, 2018) with extensive discussions about the theoretical and empirical application of the concept (Deephouse and Suchman, 2008; Suddaby et al., 2017), as well as its epistemological and ontological position (Suddaby et al., 2017). Previous comparative studies of confidence in the CPS have used a straightforward design, simply asking participants about their trust in the agency and its decision-makers (Juhasz and Skivenes, 2017; Skivenes and Benbenishty, 2022a). We use and operationalize the concepts of procedural justice that successfully have been used measuring legitimacy in the police (see Bradford and Jackson, 2016; Jackson and Bradford, 2009; Jackson et al., 2012, 2020, 2021). ¹

“Procedural justice has two key elements: first, the quality of the process used to make decisions and, second, the quality of the interpersonal treatment that people receive when dealing with authorities” (Eckhoff, 1997; Graver, 2019; Huq et al., 2011; Sankaran and Lander, 2007).

Because people’s views on child protection are an understudied area in general (Berger and Slack, 2020), and with only a few studies that compare people’s attitudes towards interventions (Berrick et al., 2019, 2021; Burns et al., 2021; Helland et al., 2020, 2022; Schmid and Benbenishty, 2011), we believe it is particularly relevant to examine and expand the knowledge base about the role and place of both people’s confidence in and moral alignment with the system. Our operationalisation of confidence is based on an understanding of procedural justice as a source of legitimacy if actions and procedures follow just laws, and power is exercised in normatively appropriate ways (Beetham, 1991 in Heywood, 2004; Rawls, 1993; Tyler and Trinkner, 2018). Ideally, procedural justice should work in such a way that also those who are disadvantaged by the outcome of a procedure still regard the decision as authoritative and binding, and that it provides sufficient reason to accept it (Sankaran and Lander, 2007; Solum, 2004). We draw out five aspects of procedural justice from the large body of existing literature: competency, fairness, respectfulness, discrimination, and moral alignment, and return to the specific survey statements in the method section.

Understanding differences in confidence level

Our study includes six European countries, and we use institutional context to understand confidence level (Berrick et al., 2021; Helland et al., 2022; Svallfors, 2012; Valarino et al., 2018). Two frameworks that relate to this idea are theories of policy feedback (which builds on the premise that policies affect politics over time; see, e.g. Béland (2010)) and policy responsiveness theory (which assumes that policymakers are aware and are incentivised by public preferences on policies; see, e.g. Brooks and Manza (2006)). A common platform for both is that people’s views and opinions on a policy or a welfare arrangement explain how policymakers and legislators orient themselves, thereby influencing how laws and policies are developed and formed. Political scientist Svallfors (2012) points out that there is a range of approaches to how welfare state research examines and applies institutional-cultural contexts and individual preferences and attitudes in empirical studies, and this multitude of approaches is also one of the reasons that it is difficult to find a unified research front (Svallfors, 2012). In our approach, we regard public opinion as a dependent variable, in which institutional context – policies and welfare institutions – influences citizens’ attitudes and their views on the role and status of welfare systems (Svallfors, 1996, 2012). We see the institutional and cultural context that people are embedded in as formative for their views on what should be a collective responsibility and how society should be built. In line with several other studies (Berrick et al., 2021; Boelens et al., 2010; Helland et al., 2022; Skivenes and Benbenishty, 2022a), we use the CPS as institutional context (see e.g. Valarino et al. (2018) using leave policy as institutional context).

The literature distinguishes between different types of CPS (Berrick et al., in press; Gilbert et al., 2011). In most European countries, one of two types of CPS is in place – namely, a risk-oriented system or a family service-oriented system. These two systems have different orientations towards families and children in need of assistance and services. In a risk-oriented system, the primary goal is to secure children’s safety, and it is a system that has a relatively high threshold for interventions and a narrow area for what falls under the state’s responsibility for action. In contrast, a family service-oriented system aims to help and support families and promotes services that aim to create change in the family circumstances. The basis for service provision and interventions is the child’s needs and the wider area for the state’s responsibilities towards children in potential need of help. Thus, the threshold here for intervention is lower than for a risk-oriented system. Family service-oriented systems also have a branch with a child-centric focus, in which children’s rights have a strong standing (Gilbert et al., 2011). Within each of the systems, there are dimensions of context including the social, political, and cultural contexts of states that shape the approach to child protection. Classifying a country into one or the other type of system depends on the orientation towards various dimensions of context and system (Berrick et al., in press), and the English system has been placed between risk-oriented and family service-oriented (Berrick et al., in press). The six countries we study represent these two systems, with England, Finland, and Norway being family service-oriented systems. Czechia, Poland and Romania represent risk-oriented systems (Berrick et al., in press; Gilbert et al., 2011; Helland, 2020). The hypothesis is that from an institutional context approach, there will be a correlation between people’s level of confidence and the type of CPS. The three countries with a family service-oriented system (Finland, Norway and England) will have the highest level of confidence, whereas those with a risk-oriented system (Czechia, Poland and Romania) will have comparatively the lowest level of confidence. Previous studies that included some of these countries empirically support the expectations for Finland, Norway and England (Juhasz and Skivenes, 2017; Skivenes and Benbenishty, 2022a).

Confidence, interventions and moral alignment

The research on the relationship between people’s level of confidence in the government and/or the welfare state (e.g. Edlund, 2006; Rothstein and Stolle, 2008) encourage us to explore a hypothesis that individuals who have confidence in the CPS would also recommend more intrusive interventions because they have confidence in the system’s ability to make correct decisions and provide help and services. Related to this is the idea that normative alignment with a system is correlated with the confidence in the system. For example, research on the police suggests that people’s normative alignment and institutional confidence is positively correlated with willingness to co-operate with the police (Jackson et al., 2020). People’s moral alignment with an institution or a system is also a central component in discussions around confidence in and the legitimacy of welfare state arrangements, requiring a normative foundation for the law, guidelines and practice that is acceptable for the people who are subjected to the regulations (Berrick et al., 2019; Rothstein, 2009; see also Svallfors, 2012). Moral alignment in child protection would concern topics related to, for example, thresholds for interventions, the support provided to families in vulnerable situations, and the balancing of children’s rights versus family or parental rights. However, a measurement of citizens’ normative alignment does not provide a measure of what they specifically mean by their alignment or non-alignment. To learn more about this we compare people’s stated alignment with the CPS with their recommendation in a specific case about potential child maltreatment.

The role of socio-demographic variables

Within welfare state research, there is no consistent pattern around a population’s socio-demographic characteristics, and primarily this is due to the wide variety of theoretical models, methodological approaches and empirical study objects (Svallfors, 2012). Although the empirical basis is small, the same appears to be the case in studies on child protection (Berrick et al., 2019; Burns et al., 2021; Helland et al., 2018). Specifically on the issue of confidence and CPS, two comparative studies have examined four and nine jurisdictions (Juhasz and Skivenes, 2017; Skivenes and Benbenishty, 2022a respectively). Juhasz and Skivenes (2017) find that gender, political orientation, income, age and education have significant effects on confidence levels. Women tend to have lower levels of confidence than men, as do those with a right-wing political orientation, those with a low income, the elderly, people living without children at home, and those without a master’s degree (Juhasz and Skivenes, 2017). The findings by Skivenes and Benbenishty (2022a) strengthen some of these findings; younger people have more confidence than older respondents, those with children have higher confidence than those without, average income level is associated with higher confidence compared with low- and high-income levels, and respondents with higher and average education level also had higher confidence compared with those with low education levels (Skivenes and Benbenishty, 2022a). Simultaneously, Skivenes and Benbenishty (2022a) do not find any significant differences in gender and marital status (live-in partners vs not), and their findings on political orientation (those with centrist political opinions have higher confidence than those with left- and right-leaning opinions) contradict those of Juhasz and Skivenes (2017). This discrepancy could be caused by variations in population, sample, measurements and how the questions were phrased, and suggests that more attention should be given to the relationship between political opinion and trust, and individual-level factors.

Method and data

This study is part of the research project “Legitimacy Challenges” funded by the Norwegian Research Council (project number 302042). The survey study has been registered in the University of Bergen’s privacy system for research, and we have submitted a Pre-Analysis Plan. ² We use an online appendix for transparency in research and supplementary material (https://discretion.uib.no/supplementary-documentation/#1552296903964-af7d19a0-9d4c). Data software will be made available to accommodate FAIR research principles (Findability, Accessibility, Interoperability, and Reuse of digital assets) after completion of the analysis (Hong et al., 2021; Lamprecht et al., 2020).

Data

The data consist of survey responses from representative samples of citizens in six countries: Norway, Finland, Poland, England, Romania and Czechia. The survey statements were developed by both authors in English. We build on previous work on citizens’ views and perceptions of the CPS as well as on the comprehensive work on trust in police and policing in the European Social Survey (ESS) Round 5 (Jackson et al., 2011). Three pilot surveys on representative samples of citizens in Norway were undertaken to test some of the preliminary questions and formulations. Recruitment of respondents and data collection were carried out by an external data collection company, Faktum Markedsanalyse. The respondents were drawn from national representative panels and the survey was issued online. ³

Faktum Markedsanalyse performed a rudimentary translation, and we undertook a meticulous check of the accuracy of the translations by engaging native-speaking research colleagues from each country to review the translations in several rounds. The survey was fielded in the six countries in the period 21–25 June, 2021. A total of 6,043 respondents completed the survey; Norway (n=1,006), Finland (n=1,004), Poland (n=1,014), Romania (n=1,001), Czechia (n=1,008), and England (n=1,010, including 864 from England, 85 from Scotland, 48 from Wales and 13 from Northern Ireland). We ran the analyses with and without the 146 respondents from Scotland, Wales and Northern Ireland, and the results showed insignificant differences. Thus, we decided to include all respondents in the analyses as England.

Measurements

Respondents were presented with five statements to measure confidence (see Table 1), ⁴ and asked to indicate how much they agree with each statement on the following scale: 1= very much disagree, 2 = disagree, 3 = agree, 4 = very much agree, 5 = do not know. The highest percentage of “do not know” responses was 26% and the lowest was 16%, with some country variation (see Table A in the Appendix for details). “I don’t know” responses are coded as missing in most of our analysis and presentation of findings, but a full overview of all responses is displayed in Appendix. For some descriptive statistics, the values for the response categories disagree (1 and 2) and agree (3 and 4) are merged (see Table B in the Appendix). The statements were presented to respondents in a randomised order to reduce memory bias and question order bias.

Table 1. Survey statements

Statement 1 measures citizens’ evaluation of child protection staff as being sufficiently competent to do their job well. Statement 2 asks about citizens’ belief in the system’s ability to make fair and impartial decisions. Statement 3 evaluates whether citizens believe that the CPS treats parents with respect and dignity. Statement 4 asks about citizens’ belief that the system discriminates against some families because of their race or ethnicity. Inherent in this statement is also the evaluation of the system’s ability to treat equal cases equally. Statements 1–4 tap into different dimensions of procedural justice (Jackson et al., 2021). Statement 5 asks citizens whether the CPS represents the same sense of right and wrong as they have – that is, do they believe that the norms and values that the CPS express through the work they do are similar or identical to citizens’ own norms and values of how the CPS should work? This is an operationalisation of normative alignment – namely, a sense that the child protection workers “act in ways that accord with societal values about how their power should be exercised” and that when they “act in ways that reflect appropriate standards of group conduct, this activates corresponding norms about how they, as citizens, should behave” (Jackson and Gau, 2015). Based on statement 5 we created two groups: those who say they are not in alignment with the CPS (responses: 1 = very much disagree, 2 = disagree), and those who say they are in alignment with the CPS (responses: 3 = agree, 4 = very much agree). The I don’t know responses are coded as missing in this analysis.

To investigate moral alignment in detail, we use respondents’ consideration of a vignette ⁵ (a fictitious but realistic scenario) about a potential child protection situation, and we asked what they would recommend the CPS to do, ranging from no intervention at all to a permanent removal of the child (see appendix Table C, see also Loen and Skivenes, in preparations). We then correlated these results with the responses on the vignette (including mean values and standard deviation). For mean calculations, we treated the recommendation response as an ordinal variable.

For some of the respondents’ socio-demographic variables, we used default questions offered by the data collection company. These are gender binary, age and age group, region of residence and income. Some of the background variables were developed by the authors and adapted from existing survey items, including size of area of residence, employment status, marital status, number of children in household, level of education and political self-reported placement on the political axis. For a full overview of the coding of background variables used in analyses, see Table E in the Appendix.

Level of confidence

We conducted a measurement validity test for the five statements to see whether they could be treated as an operationalisation of the underlying level of confidence. Cronbach’s alpha is used to determine the inter-item reliability of multiple survey items. The inter-item reliability test measures the internal reliability of the items, and whether they tap into the same concept (confidence). The standardised Cronbach’s alpha value for the five statements is 0.79, which suggests that there is moderate to high inter-item reliability (Jackman, 2008). ⁶ We also tested Cronbach’s alpha for each of the countries to check variation between countries. There is a moderate to high inter-item reliability for all six countries (see table 2).

Table 2. Cronbach’s alpha test

Note: The table shows Cronbach’s alpha values for the five item responses for each individual country and for the total sample. Data: The survey was developed by the authors and responses were collected by Faktum Markedsanalyse. RStudio: alpha-function from the RStudio psych-package (Revelle, 2021).

We construct a latent variable for confidence by taking the mean value of each respondent’s response to the five statements (Goertz, 2006). For details on the inter-item reliability and a description of the latent variable, see foot of Appendix. We use the latent variable as an expression of confidence and examine differences between countries and institutional context (CPS).

Analysis

For the descriptive statistics and OLS regression analysis, we used the statistical software RStudio (R version 4.1.1). For the significance test of descriptive percentage results, we performed pairwise t tests with p-values adjusted based on Holm (1979) multiple test procedure with sequential rejection in RStudio using the pairwise_t_test function from the rstatix package (Kassambara, 2021). To examine the impact of institutional context, pairwise t tests with p-values adjusted based on Holm’s (1979) test procedure were performed. OLS regression analysis used Finland as the country reference. To examine the correlations between confidence level (independent variable) and willingness to restrict parental freedom based on the vignette (dependent variable), we conducted a bivariate OLS analysis.

Limitations

As with all survey studies, there are limitations given that it is in general challenging to gain accurate opinions and answers of respondents in a survey. Questions and wording can be interpreted differently. Furthermore, we know that relatively few citizens have direct experience with the CPS (albeit some through indirect experience as professionals with or without mandatory referral obligations, and there is also variation across countries, see e.g. Bilson and Martin (2016)) as likely indicated by the relatively high percentage of “do not know” responses. Furthermore, responses may also reflect opinions on underlying values on family or state responsibility instead of protection of children. Panel data acquired from companies, but also in general, tend to comprise relatively fewer respondents from low-income segments. For our question on political affiliation to be indicated on a scale of 1 to 10, we marked the centre at 5, which may have skewed responses towards the value 5. In addition, we included data from the United Kingdom, which comprises England, Scotland, Wales and Northern Ireland, which admittedly have different CPS’s. However, as mentioned, there were not significant differences if we excluded the non-English responses. The CPS, in contrast to the police or the tax authorities, is a system that probably fewer citizens have knowledge about and interaction with, and thus their opinions about the system may be superficial.

Findings

Overall, about half of the populations report that they have confidence in the child protection system (51 %), 27 % do not have confidence, and about one fifth (21.85 %) responds that they do not know (see figure 1). A majority of respondents (62 %) agree on that staff is competent, and about 51-52 % agree that decisions made are fair, that parents are treated with respect and that they are in moral alignment with the system. A little less than 40 % agree that the system does not discriminate, and this is also the statement that has the highest score on “I do not know” (26 %).

Figure 1. Confidence in child protection systems.

Differences are evident between the populations in the six countries. Examining the mean values for each statement and the latent variable (Table 3 and Table D in the Appendix, including standard deviation and valid N), pairwise t tests with p-values show that England respondents score significantly higher than all the other groups in three (fairness, respectfulness and moral alignment) of the five statements, and higher than two groups (Romania and Norway) on the competency statement. Romanian respondents report a consistently low score across all five statements, and the score is significantly lower than all but the Norwegian respondents on three or more of the statements. The Czech and Polish respondents score higher than Norway on five and four statements, respectively, and Finland and England score highest. These results do not align with the expectations we had prior to the study.

Table 3. Mean level of confidence, by country

Mean level of confidence by country and total, for individual statements and latent confidence variable.

^a Mean significantly higher (p < 0.01) than Romania

^b Mean significantly higher (p < 0.01) than Norway

^c Mean significantly higher (p < 0.01) than Czechia

^d Mean significantly higher (p < 0.01) than Finland

^e Mean significantly higher (p < 0.01) than Poland

^f Mean significantly higher (p < 0.01) than England

Examining differences between the two types of CPSs (i.e. risk- or family service-oriented), the findings show that expectations are fulfilled (Table 4). Respondents in a family service-oriented system have significantly higher trust levels than those in a risk-oriented system on all statements. Citizens in a risk-oriented system have the lowest confidence on the five individual statements as well as on the latent variable.

Table 4. Mean level of confidence, by institutional context

^a Mean significantly higher (p<0.01) than Risk oriented systems

We also ran an OLS regression analysis, modelling each of the five statements and the latent variable, controlling for country-level variation, using Finland as the reference country (Table 5). In the correlation between confidence in the CPS and socio-demographic variables, four variables have some effect. Age has a negative and significant effect on fairness and respectfulness, suggesting that older individuals believe to a lesser extent than younger individuals that the CPS treats families fairly and with respect. Furthermore, older respondents and those who are politically right-oriented believe to a less extent that the CPS discriminates against some families based on ethnicity or race. Being employed and having children also have some effect in some of the models. Individuals who are employed have more confidence (fairness and moral alignment) in the system than the unemployed, and individuals with children are more morally aligned (albeit only slightly) with the CPS than people without children. For the latent variable, only education was correlated with confidence in the CPS, as individuals with higher education had more trust in the system than individuals with relatively lower education (see Table F in the Appendix for t/F-tests).

Table 5. Regression analysis of confidence levels

*** p < 0.001; ** p < 0.01; * p < 0.05.

In terms of country effects, respondents from Romania, Poland and Norway have significantly less confidence than those from Finland and England (Figure 2). The Czech respondents also have lower confidence compared with Finns, but the effect is not statistically significant.

Figure 2. Country effects on confidence levels.

There is a positive and significant correlation between confidence level (independent variable) and willingness to restrict parental freedom (dependent variable): see Table 6. We also controlled for background variables in a separate model, which indicated no notable change in the effect of the level of confidence.

Table 6. Regression analysis of willingness to restrict parental freedom

*** p < 0.001; ** p < 0.01; * p < 0.05.

Examining the recommendations for CPS intervention of individuals who state that the CPS is in alignment with their own values (alignment group), and of those who say it is not (non-alignment), the findings show overall a significant difference between the two groups (see Table 7). The responses are fairly similarly distributed on the five response alternatives and thus reflect degrees of differences. The perhaps noticeable finding is that Finland and Norway seem to stand out, as there are significant differences in the mean values between the alignment and non-alignment groups. Norway has the highest variation between these two groups, suggesting polarisation between those who believe the CPS represents their values, and those who do not. British, Czech and Polish respondents show the least variation between aligned and non-aligned respondents. For the Finns, the two groups depart on temporary placements (fewer from the non-alignment group suggest this) and monitoring (fewer from the alignment group suggest this). The Norwegians differ on several of the recommendations. The non-alignment group suggests no intervention by the CPS to a greater extent, and to a lesser extent that the child should be placed temporarily or permanently, compared with the alignment group (Table 7). The significant difference in mean values in the total sample is likely to be driven by the large sample size.

Table 7. Willingness to restrict parental freedom based on moral alignment

0 = non-alignment (Disagree: 1 and 2), 1 = alignment (Agreed 3 and 4) ***The difference in mean score between the non-alignment and alignment group is statistically significant at 99% level.

Discussion

As measured by representative samples of populations in six European countries (Czechia, Finland, Norway, Poland, Romania and England), two-thirds have confidence in their CPS. However, there are notable differences between countries, with citizens from Romania and Norway reporting lower confidence and those from Finland and England reporting greater confidence. Compared with other studies of populations’ confidence in the CPS, our findings differ on several dimensions; for example, the fact that Norwegians pair up with Romanians – and not with Finns in the upper rank. The relatively high confidence in England also deviates from previous findings. Juhasz and Skivenes (2017) and Skivenes and Benbenishty (2022a) study confidence in England, Finland and Norway, reporting that English respondents have the least confidence in the CPS, Finns show medium confidence, and the Norwegians have the highest level of confidence. These two studies specifically asked respondents about the degree of confidence they had in the decision-makers in the system (i.e. social workers and judges) and in the child protection agency. However, our results from Norway are fairly similar to previous populations studies, displaying a confidence level among Norwegians to be stable over time, even with two different ways of measuring confidence. In contrast, Finland and especially England stand out with a surprisingly high level of confidence in the CPS compared with previous results. Believing that the general level on confidence in public sector may have some similarities, we compared our results with data from ESS Round 9 (2018) and populations’ confidence in the courts and the police in which England scores lower than Finland and Norway, and above Czechia, Poland, and Romania (in this decreasing order). Although we do not have previous studies on the population’s confidence in the CPS from Romania, it seems that the findings from our study on the Romanian confidence level also stand out. The different results in these studies may be due to the survey questions. Our study asked about specific dimensions of confidence and its underlying dimensions, an approach which may result in different impact in different contexts (as Norway did not deviate from other studies).

The hypothesis of institutional context provided a clear result as confidence level is lower in risk-oriented systems and comparatively higher in family service-oriented systems. This corresponds with other population studies on child protection topics (Helland et al., 2022; Skivenes, 2021; Skivenes and Benbenishty, 2022b). These findings indicate that people have comparatively higher confidence in CPSs that have comparatively stronger authority to make interventions, and vice versa. This may be somewhat surprising as an interventionistic state with a large frontline staff inevitably will be in more situations in which it is possible to make mistakes, and thus one may think the confidence would be relatively more contested than in a system that is less “interventionistic”. However, we are here tapping into the general challenge on causal factors, as Raven et al. (2011) point out: “A major shortcoming in the existing literature on welfare state legitimacy is that it cannot explain when social policy designs follow public preferences and when public opinion follows existing policy designs and why”.

Very few notable background characteristics show a strong correlation with confidence; individuals in employment have more confidence in the system than those unemployed. We also found a weak correlation with having children. The effect of age varies on some of the confidence items: younger people believe the CPS is fairer and more respectful, and elders believe the CPS is less discriminatory. In comparative studies specifically on confidence in the CPS, the background characteristics of education level, political orientation, income level and having children correlated strongly with confidence (Juhasz and Skivenes, 2017; Skivenes and Benbenishty, 2022a). Variations are also displayed in welfare state research (Svallfors, 2012).

On the topic of associations between confidence in the CPS and recommendations for CPS interventions, our hypothesis was confirmed: individuals with higher confidence in the system also recommended more intrusive measures and were more willing to restrict parental freedom. Such findings correspond with studies of confidence and government interventions, e.g. Pitlik and Kouba (2015) documented with data from the World Values Survey the relationship between institutional confidence and confidence in government interventions. From other studies of confidence, there are findings indicating that ‘social trust’ increases support in additional education spending among elders, even though it is financed by a cutback in pension (Busemeyer and Lober, 2020). Thus, confidence can increase acceptability of welfare spending among citizens, even when the outcome does not benefit the individuals asked.

We explore the relationship between people’s view on their moral alignment with the CPS system and their recommendation for intervention in a fictitious child protection situation. Results show that although there is a significant difference between the non-alignment group and the alignment group in their recommendations, respondents are overall similar in their recommendations about what the CPS should do in the specific case they were presented with. One explanation for this may be that relatively few people in a society have direct experience with the CPS compared with the school system and the police, and thus have a less accurate knowledge about the CPS. However, two of the countries, Finland and Norway, stand out as the non-alignment group clearly recommends less intrusive interventions compared with the alignment group. Both these countries have a family service-orientation with a child-centric focus (Gilbert et al., 2011), and we believe this create tensions: a child-centric focus requires the state to regard the child as an individual and emphasises and protects the child’s individual rights and, if necessary, irrespective of parental rights and the family interest. Whereas with a family-service orientation, the state interacts with the family as a unit and to a lesser degree would restrict family autonomy. In most countries with an established CPS this dilemma will be prevalent, but in countries in which there is an increasing child-centric focus we anticipate that there is a stronger potential for tensions around conflicting principles (i.e. family privacy and individual child rights). It may be that citizens in these two societies note that the government has taken upon itself to protect children’s rights – even if this is at the expense of the family unit – and thus there is evident polarisation around core values in these two societies.

Concluding remarks

Our study contributes to several ongoing streams of policy discussions. First, the literature on legitimacy and confidence in welfare states has so far very little research on citizens’ view on child protection and children’s rights. Our study provides valuable insights into citizens’ confidence in relation to the CPS which increases our understanding of the legitimacy of the state, public administration, and the judiciary. These state interventions have the potential for creating public outcry and protests (Skivenes, 2021). A reason for this is that the state’s responsibility for children and intervention in families on behalf of children questions parenthood and family – core institutions in society (Berrick et al., in press; Gilbert et al., 2011). Children and parents have an extensive set of rights, including a right to family life, and children are both independent and dependent actors. Child protection interventions can be extremely consequential for parents and children and are often contested and criticised. On the one hand because there is legitimate disagreement around how to interpret the value-laden principle of the best interest of the child, and the thresholds for what is a sufficiently good upbringing of children. On the other hand, interventions may be contested because child protection decisions often are based on insecure and partial evidence due to the private charter of the family sphere, and decisions must include impossible prediction of the impact of an intervention on the child’s future health and happiness. These factors increase the importance of knowledge about legitimacy dimensions for governments and policy formation.

Second, this study contributes with some new insights into the literature on institutional context and expands our knowledge base on the impact of established authorities, which is an interesting branch in policy research (see, for example, Rothstein, 2009; Svallfors, 2012; Valarino et al., 2018). Thirdly, the paper adds to the knowledge between persons beliefs on their moral alignment and their recommendation for state intervention, which complement the findings on institutional context. To our knowledge, no previous study has examined the beliefs and viewpoints of individuals who believe that the CPS does not operate according to their moral compass (i.e. non-alignment) or vice versa (alignment). It is interesting that there seems to be a tension about the recommendations of state interventions in the two societies that are known to be the most child-centric – namely, Finland and Norway. This may indicate some sort of legitimacy challenges in this area of the welfare state, and we believe it would be important to follow up with additional studies of polarisation tendencies and variations around views on state interventions in the family.