Parental Freedom in the Context of Risk to the Child: Citizens’ Views of Child Protection and the State in the US and Norway

Abstract

Child protection is considered an appropriate government responsibility, but interventions into the family are also some of the most consequential for states. This study examines the normative basis for limiting parents’ freedom by exploring public attitudes about a child’s safety in the context of increasing risk. Using a randomized survey, we test the causal relationship between levels of risk and parental restrictions on representative samples in Norway and CA, US (n = 2148) – different welfare state and child protection models. Findings suggest that the public supports restricting parental freedom under conditions of risk and that severity of risk is taken into consideration. A majority favour restricting parental freedom under conditions of risk to the child; a minority resist restricting freedom, regardless of risk, and about one-third to one-half of respondents favour temporarily suspending parents’ rights by separating children to foster care. Residents of Norway are half as likely to support unrestricted parenting, regardless of risk, and are 1.5 as likely to endorse restricted parenting. Norwegians are also 20% less likely to support separating a child from his parent compared to US respondents. The study has implications for system design based on popular notions about parents’ freedom and family privacy.

Introduction

All western industrialized countries place some limitations on individual freedom and governments usually devise policies to limit parents’ absolute freedom when their actions result in harms to children or youth (Berrick et al., in press). When a child’s safety or well-being is seriously compromised, the state can impose certain parental restrictions, assume temporary parental responsibility, or in extreme circumstances and in some countries, it can terminate all parental rights if parents are unable or unwilling to perform their parental commitments. To secure children’s safety or their best interests, the state might require parental compliance with services, or place children in foster homes or residential units; or it could allow another family to adopt the child. These laws and practices, along with the offer of voluntary family supports, typically represent some of the essential elements of the child protection (variously called child welfare) system of a given state (Gilbert et al., 2011). But institutional frames across countries differ. As an example, some states offer a wider or richer array of primary prevention services, and others may be more reliant on secondary or tertiary intervention (Berrick et al., in press).

These state interventions in family life represent an immensely strong state power. Child protection decisions simultaneously challenge parental freedom and the privacy and autonomy of family life (Shapiro, 1999). Decisions that result in state intervention, therefore, should – at least to some degree – reflect within-country societal views about children, their safety, and the threshold of risk that can be tolerated. Societal views about children, of course, are always undergoing change. Some behaviors toward children that would have been normatively accepted 100 years ago are no longer appropriate today. Whether these changing perspectives have influenced the development of child protection policy, or whether changes in policy have brought about changes in public attitudes are contested notions (Béland, 2010; Brooks and Manza, 2006). What is known, however, is that child protection is harshly criticised in the mass-media when social workers and judges – as agents and arbiters of the state – make determinations about children’s risk that might inappropriately limit parents’ freedom, or when they inaccurately underestimate risk and children are harmed (Biesel et al., 2020). Where the public sets its threshold of tolerance for certain parental behaviors, and where the state chooses to intervene may or may not align. In this study, we examine the public’s views about acceptable limits on individuals’ liberty and if, when, and how governments should intervene in family life. This observational study uses a survey design to test the causal connection between risk level and degree of state restriction, and includes a representative sample of the population (n=2148) in two countries. Our hypotheses are first that increasing risk to a child will be associated with increasing acceptance of restrictions on parents, and second, that we will find differences between institutional contexts. The study has implications for child protection policy design in unique country contexts.

The paper starts with its theoretical platform, followed by an examination of restrictions on parental liberties in California (USA) and Norway relative to child protection policy. We present the methods for the study, followed by findings and a review of the implications of these data for child protection policy.

Theoretical approach – freedom and institutional context

Basic human principles lie at the heart of child protection policy, including values related to the state’s relationship to vulnerable families and how children should be raised by their parents. Values relating to care for children are universal (LeVine, 1988), but parenting strategies and culturally-based principles can be quite dissimilar in different country contexts (Faircloth et al., 2013). Little is known about how the population considers the value of parental freedom in the context of children’s safety and risk, yet the design of public child protection systems is fundamentally based on normative values about children’s safety and risks, and parental freedom and autonomy. These beliefs are socially constructed, historically bounded, and mutable with time and changing cultural mores. It is not clear from the literature how institutional context with its societal values and beliefs are connected to attitudes about child protection policies (Berrick et al., in press; Parton, 2019). This study is rooted in comparative child protection attitudinal research and aims to increase our understanding of the mechanisms and factors that influence views about and acceptance of state intervention policies that protect children’s well-being interests and limit parents’ liberty interests.

All states place some limitations on individual freedom, but in the field of child protection, the liberty interests of the child to be free from abuse or neglect sometimes compete with the interests of parents to freely care for their child as they would prefer (Holland and Scourfield, 2004). Inappropriate infringements on either party’s liberty interests raise questions of justice and must therefore be considered carefully. Philosopher John Rawls’ (1999) theory of justice describes fundamental principles of personal liberty: “Each person is to have an equal right to the most extensive scheme of equal basic liberties compatible with a similar scheme of liberties for others.” (p. 53). Rawls’ theory, applied to the field of child protection, would therefore suggest that a parent’s liberty interests to care for and/or treat their child as they choose is constrained by the child’s rights to liberty. How these competing liberty principles are resolved and where different country contexts set the boundaries of their resolution have not previously been well articulated. The famous anti-paternalist principle of John Stuart Mill states that harm is the only just reason for restricting freedom:

The only purpose for which power can be rightfully exercised over any member of a civilized community, against his will, is to prevent harm to others. His own good, either physical or moral, is not a sufficient warrant’’ (Mill, 1859, p. 78).

In child protection, harm or risk of harm to a child offers the justification for intervention, though different states may narrowly or broadly define harm to include physical, mental, developmental, or other aspects of well-being.

In political theorist Isaiah Berlin’s conceptualization of liberty (Berlin, 1969 [2017]), he suggests that citizens’ political relationship to the state may be fundamentally characterized as negative or positive. In child protection, Holland and Scourfield (2004) show how Berlin’s notion of negative liberty applies. Negative liberty refers to the freedom that comes with few external constraints or freedom from government intrusion into family life. In this sense, Berlin’s views relating to negative liberty are not notably dissimilar to Mill’s libertarian sentiments. A child protection system designed around notions of negative liberty would allow for a wide range of parenting liberties to be expressed absent state intervention to prevent all but extreme behaviors. Positive liberty, on the other hand, refers to the opportunities among individuals to act unfettered, or their freedom to. Berlin introduces the idea of positive liberties thus:

The ‘positive’ ‘sense of the word ‘liberty’ derives from the wish on the part of the individual to be his own master. I wish my life and decisions to depend on myself, not on external forces of whatever kind. I wish to be the instrument of my own, not of other men’s (sic), acts of will. I wish to be a subject, not an object; to be moved by reasons, by conscious purposes, which are my own, not by causes which affect me, as it were, from outside (Berlin, 1969 [2017], p. 178)

In this sense, political structures that enhance opportunities to realize one’s hoped-for capacities would be liberatory. Child protection efforts would be oriented around these positive rights if they support parents’ capacities to be self-directed and realize their best parenting selves as well as children’s life chances,.

In the case of international comparative child protection approaches, the US (and therefore California) child protection system is heavily weighted toward a negative liberty paradigm. Gilbert and associates (Gilbert, 1997; Gilbert et al., 2011) have characterized the US as a child protection system that is risk-oriented, where the state tolerates a relatively high threshold of risk or harm before the liberty interests of the parent are inhibited or restrained. Parents have wide discretion to care for their children unrestrained from the government. By way of illustration, in the 1920s, two important US Supreme Court decisions privileged the rights of parents to make educational decisions for their children in contrast to the state’s interest in children’s education (see: Meyer v. Nebraska, 1923; Pierce v Society of Sisters, 1925). A number of US Supreme Court cases since then have tested the limits of parental freedom only to repeatedly underscore parents’ rights “in matters of family life” as a “fundamental liberty” (see Santosky v Kramer, 1982). Guggenheim has noted several instances when the US Supreme Court has signaled the importance of a parental rights doctrine. In Stanley v. Illinois, the court stated, “the interest of parents in the care, custody, and control of their children” is “perhaps the oldest of the fundamental liberty interests recognized by this court.” And in Troxel v. Granville, the court indicated parents’ rights to raise their children as “essential”, and “rights far more precious… than property rights” (Guggenheim, 2005: 18). Although the word “parent” is absent from the text of the US Constitution (as is the word “child” or “children”), at least four constitutional amendments undergird the privileges of parental liberty. These include the 1^st, 5^th, 9^th, and 14^th amendments (Skinner and Kohler, 2002). In particular, the Equal Protection Clause of the 14^th amendment suggests that tests of the parental liberty principle must be subject to “strict scrutiny” and state efforts to override parental freedom must be “compelling” and must be exercised using the “least restrictive means” (Skinner and Kohler, 2002). The Supreme Court has also determined, however, that in certain circumstances, parental freedom is not limitless. On issues pertaining to child labor, vaccination, or school attendance, for example, parental freedom may be constrained. Relevant to child protection, parental liberty may also face limits in circumstances of child endangerment (see Prince v. Massachusetts, 1944).

In contrast to the US, the Norwegian constitution has adopted and incorporated a number of human rights conventions promulgated by the United Nations. These conventions stipulate a range of positive liberties which oblige the state to support and protect individuals. All individuals – including children – have civil, political and social rights. Although parental rights have strong standing and the privacy of the family sphere is protected, compared to the US, the Norwegian system allows for far more state intervention oriented to support positive liberties. According to Gilbert et al. (2011), Norway’s child protection system is characterized as oriented toward a “family support” perspective. With a rich array of supportive services offered universally and voluntarily to all families, and long-lasting, saturated services offered to targeted families, the Norwegian child protection model aspires to Berlin’s notion of “positive liberty”, where the individual is afforded opportunities to be free to: to be supported to become the ideal-type or at least a better parent.

Just as parents enjoy positive liberties, so too do children in Norway. The UN Convention on the Rights of the Child (UNCRC) was ratified early on and in 2002 was incorporated into all Norwegian legislation. In 2014 the Constitution was amended to include a new article (§104) about children’s rights:

Children have the right to respect for their human dignity. They have the right to be heard in questions that concern them, and due weight shall be attached to their views in accordance with their age and development. For actions and decisions that affect children, the best interests of the child shall be a fundamental consideration. Children have the right to protection of their personal integrity.

The authorities of the state shall create conditions that facilitate the child’s development, including ensuring that the child is provided with the necessary economic, social and health security, preferably within their own family.

As a result, the state is obliged to protect children’s liberty interests co-equally with parents’ liberty interests. With this strong child rights orientation, the state has a direct responsibility for protecting the child’s interests, shared with the parent’s obligation to the child.

The US and Norway, with their distinctive institutional contexts of family policy and child protection, are nested in welfare states that are typically portrayed as quite different from one another (cf. Healy and Oltedal, 2010). Esping-Andersen’s (1990) welfare state regimes place Norway solidly among the social democratic welfare states, and the US among the liberal welfare states, the former offering a welfare safety net that strongly buffers adults from the uncertainty of the market. The US (and California) has a relatively thin social safety net with few services, where the individual is largely responsible for assuming risks associated with modern capitalism (Hacker, 2019). Perhaps related, the economic conditions of individuals and families in the two countries are notably different. The Gini coefficient (a measure of income inequality where higher values represent greater income inequality) in the US is 45 (49 in California) (estimated in 2007) compared to 26.8 (estimated in 2010) in Norway (Index Mundi, 2020). Each of these welfare systems are ideologically distinctive in their approach to social responsibility, social problems and social inequality overall, and the degree to which the state is involved in families’ lives.

A review of some of these country context differences is provided in Table 1. These distinctive characteristics make international comparative analyses especially fruitful as they highlight unique policy choices available in response to similar child and family difficulties. Moreover, comparison of Norway and the US (California) is appropriate as these countries share a number of common features. For example, both are highly developed western industrialized nations, and are ranked as high-income countries. Less than 25% of the total population is made up of children (U.S. 22.3%; CA 22%; Norway 23.8 [Statistisk sentralbyra, 2021; U.S. Census, 2021]) and the population growth rate in both countries hovers below 1% (2019 rate: U.S. 0.5%; CA 0.0%; Norway 0.7% [MacroTrends, 2020; World Bank, 2019]). Both countries have a life expectancy that is roughly similar at 82.2 (Norway), 78.7 (U.S.), and 80.8 (California; Arias et al., 2021; Statista, 2020). The population of Norway is predominantly white and Norwegian (about 83%) and the large majority of the US population (76.3%) is white, though there is greater racial diversity in California (55.8% white; Index Mundi, 2020; U.S. Census, 2021).

TABLE 1. Selected country context characteristics, California, USA and Norway

Although the policy contexts differ, there is a knowledge gap regarding how the public views children at risk, the threshold for intrusive intervention, and the degree to which parents’ freedom should be restricted. In this study we build upon prior work that examined public attitudes about the provision of state-sponsored services in a situation of low-risk to the child (Berrick et al., 2020). Here we solicit public attitudes about parental freedom as it relates to child rearing. We use an observational, experimental study to examine three levels of risk to determine the threshold when public attitudes shift. We further examine these public attitudes towards risky child rearing in different country contexts where the child protection policy paradigm notably differs. And we include an analysis of demographic variables to determine whether underlying population dynamics or country context are more prominent in our explanation of outcomes. Specifically, we address two hypotheses. The first focuses on values relating to parental freedom, and the second on institutional context: H1. Citizens will tolerate greater restrictions on parental freedom as a child’s risk rises. H2. Norwegians will have greater tolerance for restrictions on parental freedom than residents of California (USA).

Methods

Sample

This observational study uses a survey vignette distributed in two countries – Norway and the US – designed to examine citizens’ views about government intervention and restrictions on parents’ freedom. The sample includes 1,031 respondents from Norway and 1,117 respondents from one US state (California). US federal policy provides a general frame and funding for the organization of child protection systems, but prior research shows significant variability in child welfare policy and practice across the 50 US states and territories (see for example, Edwards, 2016). Selection from one state is therefore an appropriate recognition of the variability across states, and allows for a sample size that can serve as representative of that state. California also represents about one-fifth of the entire US population, and is therefore highly consequential for understanding a significant portion of US public opinion and child protection practice.

We used a data collection firm in Norway (Respons Analyse [RA]) and in California (YouGov) to collect the data. RA maintains a representative sample of Norwegian adult residents as potential web-based survey respondents as does YouGov in California. This panel is regularly employed by RA to answer questions relating to a wide range of topics including brand measurements, attitudes, and behaviors. The respondents are representative of the populations of California and Norway, respectively, through a weighting procedure. ¹ The required human subjects’ permissions were granted from the authors’ university institutions.

Instrument

A survey vignette was used (Wilks, 2004) to assess citizens’ attitudes about children, parents, and state interventions into the family. Vignettes have been used successfully elsewhere as a strategy to compare and understand underlying values across different country contexts (e.g. see: Benbenishty et al., 2003; Skivenes and Tefre, 2012; Soydan, 1996). The vignette characterizes a vulnerable family with implications for child maltreatment. In the survey, we manipulate one variable that represents the severity of risk to the child, which is established by age (i.e. an infant) and condition (i.e. born addicted to drugs due to serious parental drug use). The vignette reads as follows:

A social worker visits Julie in the hospital when Julie gives birth to a baby boy. Julie is addicted to drugs and the newborn is suffering from drug withdrawal symptoms. The social worker is very concerned about the baby’s safety, assesses Julie’s ability to take care of the baby, and recommends drug treatment for Julie. Julie says she is sorry that she may have hurt her baby, she realizes she has a serious problem, and she is willing to enroll in treatment.

One sentence in the vignette, underscored, is manipulated to characterize severity of risk conceptualized as parental cooperation, insight, and responsibility. X1 is low risk (as it reads in the vignette above) because it represents high levels of cooperation, high insight, and high levels of responsibility. Medium risk (X2) is represented by medium levels of parental cooperation, medium insight into the problem, and medium levels of responsibility: “Julie says she is not sure she may have hurt her baby, she thinks she may have a small problem, and she doesn’t need treatment.” The high risk context (X3) – “Julie says she did not hurt her baby, she does not think she has a problem, and she refuses to enroll in treatment” – represents low levels of parental cooperation, low insight into the problem, and low levels of responsibility. By varying the independent variable, we can examine if level of risk is causally related to the dependent variable, restricted freedom. The vignette and accompanying survey and demographic questions were developed in American English then translated and back-translated into Norwegian. The instrument was assessed for face validity by researchers and child welfare practitioners in both countries.

Respondents were randomly assigned a vignette with either X1 (n=701), X2 (n=700), or X3 (n=747). The three treatment groups of respondents are overall similar on sociodemographic variables as displayed in Table A1, Appendix A.

Respondents’ attitudes about restrictions on parents’ freedom and the appropriate role of the state are assessed with three statements that serve to measure increasing restrictions: Unrestricted parenting – “Julie should be free to bring her baby home regardless of the social worker’s assessment.” Restricted parenting – “The baby should stay with Julie in a supervised setting.” Suspended parenting – “The baby should be placed in foster care.”

Respondents were given a 4-point Likert scale from Strongly Disagree (1) to Strongly Agree (4) for each of the three statements. The approach reflects some of the complexities found in child protection cases and the choices decision makers face; respondents are given the opportunity to consider each response option. We see a response statement of unrestricted parenting as mutually exclusive from restricted parenting and suspended parenting. Both restricted and suspended parenting responses reflect different forms of restricted parenting with suspended parenting the most invasive.

Background variables relating to respondent gender, metropolitan area, job status, political orientation, domestic partner status, education level, religion, immigration status, income level, age, and head of household with children were included to determine if these might mediate findings relating to the dependent variable.

In the Appendix, Table A1, we provide an overview of the sample, by country and risk severity. In Table A2, we provide an overview of mean values, standard error, and n for each treatment and response, in total and per country. Figure A1 displays corresponding mean values. We also performed a supplementary analysis in which we merged values 1 and 2 into a disagree category, and 3 and 4 into an agree category for each treatment (see Tables A3-A5). Supplementary findings from our mediation analysis are displayed in Table A6.

Analysis

The statistical program Stata SE Version 15 was used for data analysis. An omnibus ANOVA test is used to determine whether there is an overall statistically significant treatment effect, and post-hoc Bonferroni-corrected multiple comparison tests are used to test for significant differences between mean values. We report significant differences at p<.01 (**) and p<.001 (***).

Next, mediation between country and parental restriction was examined using the R structural equation modelling package Lavaan version 0.6-6. Single mediator models were constructed to measure the indirect effects of 11 potential mediators: (1) gender (0=male, 1=female), (2) metropolitan area (0=small (<100,000 inhabitants), 1=large), (3) job status (0=unemployed, 1=employed – Respondents were coded as “employed” if they identified current job status as “employed full-time”, “employed part-time”, “self-employed/freelancer”, or “parental leave”. Respondents were coded as “unemployed” if they identified current job status as “student”, “retired”, “pensioner”, “jobseeker”, “homemaker”, “incapable of work due to disability”, or “income solely from public assistance”.), (4) political orientation (0=least conservative, 2=most conservative), (5) domestic partner status (0=no partner, 1=partner), (6) education level (0=high school diploma or less, 1=any college education, 2=graduate degree), (7) religion (0=not religious, 1=religious), (8) immigration status (0=non-migrant, 1=first or second generation immigrant), (9) income level (0=up to US$49,9999, 1=up to $99,999, 2=up to $500,000), (10) age (0=18-34, 1=35-54, 2= 55+), (11) children (0=no children in the household, 1=one or more children in the household).

We also calculate the percent of the total effect that each mediator explains. When more than one indirect effect was significant for a given level of parental restriction, a multiple mediator model was constructed to simultaneously measure the indirect effects of the mediators while accounting for their covariance (see Table A6 in the Appendix). Non-parametric bootstrapping was used to measure standard errors of effects (replications=1,000).

Findings

Findings from this study suggest that our first hypothesis is largely confirmed. As risk to the child rises, the public tolerates greater restrictions on parental freedom (see Table 2 below, and Table A2 in the Appendix). The omnibus ANOVA tests suggest that respondents are less likely to support unrestricted parenting as risk rises (p<.001), and they are more likely to agree with suspended parenting as risk rises (p<.001). Results regarding restricted parenting do not follow the anticipated risk trajectory; as risk rises, respondents are more likely to disagree with restricted parenting (p<.01).

TABLE 2. Mean values and treatment effects on views of parenting restrictions by severity of risk, overall and by country (n=2,148). 1=strongly disagree, 4=strongly agree. (Omnibus ANOVA and post-hoc Bonferroni-corrected multiple comparison tests).

** omnibus ANOVA significant at p < 0.01

^l multiple comparison test with low-risk level significant at p<0.01

^m multiple comparison test with med-risk level significant at p<0.01

^h multiple comparison test with high-risk level significant at p<0.01

The treatment effect is also valid for the within-country samples with regard to unrestricted parenting and suspended parenting. As risk to the child rises, Norwegians and California respondents are less likely to endorse unrestricted parenting, and more likely to endorse suspended parenting. Norwegians are less likely to agree with restricted parenting as risk rises in this scenario. We did not see differences with regard to restricted parenting in the California sample (see Table 2).

Merging responses “agree or strongly agree” and “disagree or strongly disagree” into binary options, we find that the majority of respondents (75-81%) favoured a restricted parenting approach, across risk groups. Fewer than one in five respondents agreed with a state response that allowed unrestricted parenting. About 20% (19.3%) agreed with unrestricted parental freedom in the low-risk context, followed by 15.3% in the medium risk context, and 12.4% in the high risk context (see Table 3 below as well as Tables A3-A5 in the Appendix). When examining responses in the total sample to our measure of suspended parenting, we see that less than half of respondents agree with a state action that would result in separating the parent and child. About one-third of respondents (34.5%) agreed with separating the parent and child in circumstances of low-risk, about two of five respondents (42.7%) supported parent-child separation under circumstances of medium risk, and in the high-risk situation, almost half (47.9%) of respondents supported suspending parental freedom.

TABLE 3. Percent “Agree” by level of risk and parental restrictions

Note. p-value denotes significance of test of proportions between Norway and California

† p ≤ 0.006 (Bonferroni-corrected p-value)

The second hypothesis is also confirmed. There are significant differences between Norwegian and California public attitudes about restricting a parent’s freedom. Across all statements and risk levels (except for views about suspended parenting under conditions of medium risk), Norwegians hold more favourable views than Californians about restricting parental freedom (see Table 2 and 3). Regardless of risk, respondents from Norway are less supportive of unrestricted parenting than respondents from California and they are more supportive of restricted parenting. Except for the medium risk scenario, respondents in Norway are clearly less supportive of suspended parenting than residents of California. As risk rises for the baby in this vignette, Californians’ support for suspended parenting increases.

We find that less than one in ten Norwegian respondents (8%, 7.4%, and 4.9% depending on risk from low to high), agreed with unrestricted parenting compared to 29.6%, 22%, and 19.8% respectively in California. Over four-fifths of Norwegian respondents (87%-low, 85.1%-medium, and 81.3%-high) agreed that the circumstances warranted restricting Julie’s parenting, compared to about three-fourths of California respondents (76%-low, 74.5%-medium, and 68.3%-high). And there was also support for suspended parenting – a measure of placement into foster care. Twenty-eight percent of Norwegians supported suspended parenting in the low-risk situation compared to 42.7% in medium risk, and 40.9% in high risk. In California, 40.3% supported suspended parenting in the low-risk situation compared to 42.7% in medium-risk, and 54.8% in the high-risk circumstance (see Table 3).

Mediation analysis

Examining if background variables mediate the effect of country, the results from single mediator models (see Table 4) show that respondent age was a significant mediator of the association between country and parental restriction for all three levels of restriction. Norway respondents were half as likely as California respondents to agree with unrestricted parenting (OR=0.46, p<=0.001), and nine percent of this effect was because Norway respondents were more likely to be older, and older respondents were more likely to disagree with unrestricted parenting. Norway respondents were 1.5 times as likely as California respondents to agree with restricted parenting (OR=1.49, p<=0.001). However, older respondents were less likely to agree with restricted parenting, and because Norway had a greater proportion of older respondents, the indirect effect (OR=0.96, p<=0.01) suppressed the direct effect (OR=1.55, p<=0.001). Norway respondents were 20% less likely than California respondents to agree with suspended parenting (OR=0.80, p<=0.001). Nineteen percent of this total effect was because Norway respondents were more likely to be older, and older respondents were more likely to disagree with suspended parenting.

TABLE 4. Mediation of Association Between Country and Parental Restriction

Notes. Bootstrapped standard errors (reps=1,000)

** p≤0.01

*** p≤0.001

Migration status (see Table 4) was also a significant mediator for the association between country and unrestricted parenting (OR=0.87, p=0.001). Norway respondents were less likely to be immigrants, and immigrants were more likely to agree with unrestricted parenting. Immigration status explained 18% of why Norway respondents were half as likely as California respondents to agree with unrestricted parenting. In the multiple mediator model of age and immigration status (see Table A6 in the Appendix), both remained significant mediators.

Discussion

This study included a representative sample of residents from Norway and California (US) to elicit their views about children at risk and the responsibilities of the state to restrict parenting under conditions of variable risk. By randomly varying the independent variable, the current study extends our capacity to understand how or if child welfare policy contexts are associated with public attitudes, particularly with regard to weighty decisions associated with restrictions on parental freedom.

We have largely confirmed both of our hypotheses. We find a clear treatment effect associated with the severity of risk. As risk rises, there is a commensurate rise in willingness to restrict parental freedom. There is general approval for restricted parenting regardless of severity of risk to the child. Furthermore, we find differences in respondents’ views based on institutional context; Norwegian residents are more likely to favor restricting parental liberties compared to residents of California, and California residents are more likely to support unrestricted parenting. One third to more than half of respondents from both groups are willing to suspend parental liberties through foster care, with California respondents viewing foster care more favorably as risk rises. Our findings indicate that age is a mediating factor for institutional context as an explanatory variable, though these findings are difficult to interpret. Older participants in Norway offer responses notably different from other respondents. They are less likely to favor any of the three alternatives (i.e. unrestricted parenting; restricted parenting; suspended parenting). Trust in government might account for this, as one study showed that older respondents have less trust in child protection systems (Juhasz and Skivenes, 2016). Other authors have examined citizens’ attitudes toward the welfare state, in general, and in the context of expanding or contracting welfare states over time (Blekesaune and Quadagno, 2003; Laenen et al., 2020; Svallfors, 2012). These studies suggest there is wide variability in public attitudes across states and that individual values and personal circumstances play a role in shaping citizens’ views. In general, age is not strongly determinative of public attitudes toward the welfare state (Edlund and Svallfors, 2011; Busemeyer et al., 2009); age-related findings from this study therefore do not fall neatly into the larger welfare state literature.

Mill’s harm principle (1859) – that government can only restrict freedom under conditions of harm to others – may be at play in respondents’ attitudes about restricted parenting. The principle is considered a legitimate reason for state intervention in many countries; however, it may have different interpretations depending on individual or societal contexts. Respondents from California who favored unrestricted parenting may have embraced the freedom aspect of Mill’s principle more fully and/or assumed a higher harm standard than Norwegian respondents. Notable is the relatively large percentage of California respondents (between 20-30%) who favored unrestricted parenting, regardless of risk. We see this finding as emblematic of the negative liberty perspective offered by Berlin (1969 [2017]), and suggest this as a relatively strong signal regarding the public’s views of the parents’ rights doctrine that prevails in the US. This finding may also reflect a general skepticism toward government in the US. Findings from other studies indicate that public confidence in the child welfare system is not particularly high; public attitudes about child welfare in the US and the UK, in particular, are more negative than in the Nordic countries (Juhasz and Skivenes, 2016). Findings regarding Norwegian migrants’ more positive views about unrestricted parenting may reflect a cultural view on parenting or suspicions about government intervention. Further study of this issue is warranted.

In contrast to the US, Berlin’s principle of positive liberty may have been influential in Norwegian responses. In that regard, when children’s health or safety are at risk, restrictions on parental freedom allow government agents to provide children with the goods, skills and abilities necessary to become autonomous adults. Restriction on parents’ liberties and a concomittant requirement to engage in services may also reflect a positive liberty approach to enhancing parents’ capacities to fully engage in positive parenting practices with their children. Using government as a tool to optimize opportunities and capacities, those who favor parental restrictions may see government services in a positive liberty frame.

Furthermore, responses may have been rooted in a general understanding of the child welfare frame in which citizens live. A large majority recommended restricting parental freedom in the vignette provided, regardless of the parent’s compliance or insight. In Norway, respondents’ views largely reflect the child protection system available to families. Services, where mother and baby might live together for some time, would likely be available and possibly required. Moreover, the emphasis on child rights in Norway, as an early adopter of the UNCRC, and embedded in the national constitution, might suggest that survey participants would be attuned to children’s liberty needs at the expense of some restrictions on parents’ freedom.

In the US, studies suggest that substance abuse is often a contributing factor to child maltreatment (US DHHS, 2019; Walsh et al., 2003). Although substance abuse would not necessarily be determinative of child protection involvement, the data suggest it can be an important, related factor. California respondents’ views about restricted parenting, less favorable than Norwegian respondents but still high, may signal acknowledgement about the risks associated with substance abuse, particularly for infants. Given the US ethos of individual responsibility (Prager, 1994), and the high risk scenario in which the parent did not accept responsibility for the child’s safety, this aspect of the case scenario may have influenced respondents’ views.

One-third to over one-half of respondents expressed a willingness to separate the baby from the mother – suspended parenting – and California respondents were more likely to favor this option with increased risk to the child. Why are there consistently high proportions of California respondents that approve of suspended parenting when they are otherwise less willing than the Norwegians to restrict parental freedom? Foster care serves as an extreme form of state intervention and as such, these results are surprising in California, given the negative liberty frame of American governance. A possible explanation for these somewhat contradictory results may be that both Berlin’s negative liberty principle and Mills’ harm principle may be at play and that these ideas are very much in competition with one another in child protection, where Berlin’s principle might prevail with regard to parents, but Mills’ principle might be taken into account with regard to the child.

Other research indicates that regardless of one’s role – as community member, social worker, or judicial actor – and one’s country of residence, decisions about parent-child separation are weighty and are typically reserved for serious circumstances (Berrick et al., 2020). Nevertheless, California respondents may have been reflecting their understanding of typical child protection services. Critics of the US child protection system have long argued that the service system is thin, with few services other than foster care available to respond to children’s needs (Roberts, 2009).

Limitations

This study provides new and important insights into the public’s views about children, their safety, and the role of the state in protecting children by limiting parental freedom. There are limitations. The vignette only captured views relating to infants and to a particular safety threat, limiting the general empirical value of our findings. The vignette also does not refer to social class or to other characteristics of families that might shape respondents’ views. Our operationalization of increasing intrusions, from unrestricted to restricted to suspended parenting, although reflecting the complexity faced in these cases, is not an exact scale. This may have obscured some of our results. Our survey may also hold certain biases to which we are unaware. The sample included in this study is representative of the populations in Norway and in California on a given set of key variables; as in any public opinion poll, however, we cannot check the veracity of respondents’ stated views. The data from the US are derived from a California sample. Although the US has been characterized as a liberal welfare state and is thus significantly different from Norway, a large body of research also suggests that there are important regional variations across the US states and that California is more representative of other Western and Pacific states than it is representative of the country as a whole in its politics, personality, and social correlates (Rentfrow et al., 2013; Woodard, 2012). Nevertheless, California’s child protection system is heavily reliant on federal funding and policy guidance, so the inclusion of a single state is nonetheless instructive about the larger federal context. Certain details are also omitted from our analysis of respondents. For example, we included a measure of “employment” to determine whether a respondent was currently participating in the labor market or not. A more fine grained examination of respondent characteristics might include detail about one’s type of employment since individuals participating in some employment settings might view child protection responses differently. Lastly, although immigrantion status was found to be a significant mediator of the observed association between country and unrestricted parenting, there were relatively few immigrants in the Norway sample (n=47); thus these findings should be interpreted cautiously.

Conclusion

The Declaration of Independence, the Constitution, the Bill of Rights and subsequent amendments provide the symbolic and legal frame for public policy in the US. Each of these documents identifies specific inalienable rights that the government must protect, including negative liberties such as the right to freedom from government interference. Findings from this study suggest that even under conditions of considerable risk to a child, about one-quarter of the California sample believed that parents should be unfettered in their care for children. In contrast, the Norwegian sample’s tolerance for state restrictions on parental liberty is much greater. With a constitutional frame that guarantees children have rights that are equal to those of other citizens, along with a state welfare infrastructure that provides a relatively thick web of protection for citizens, Norwegians are highly unlikely to view parenting through an unrestricted lens. Under conditions of risk, Norwegian residents largely support state involvement in family life.

A majority of respondents in both countries favored restrictions on parental freedom, particularly as risk to the child increased. One restriction presented to respondents was a “supervised setting” in which the parent can care for their infant child safely. In Norway, such settings are generally available, as needed. In the US, by contrast, the bulk of federal child welfare funding (the Title IV-E) is dedicated to covering the cost of care for children who have been separated from their parents; co-resident services are rare. Recent US legislation, the Family First Prevention Services Act (2018), will expand access to federal funding for services to children who are “at risk” of foster care placement, including substance abuse services deemed “promising, supported, or well-supported” by research evidence. Although the kinds of co-residential substance abuse service programs alluded to in this study’s vignette are not inexpensive, we anticipate that they will become more widely available and used in the newly emerging Family First policy context. In this regard, US policy will soon be considered more responsive to the values held by the general public, making available services that restrict parental freedom, rather than suspending their freedom altogether.

To the degree that national child protection policies reflect commonly held public values and attitudes (cf. Heimer et al., 2018), these alignments are likely to garner greater public support. The child protection policy context in Norway appears to be largely responsive to public attitudes about limits on parental liberty when children’s safety is at stake. Findings from this study also suggest that policy making in the US in the field of child protection may be more difficult because respondents’ attitudes are more diverse. Striking an appropriate balance between the liberty interests of parents and children is the fundamental challenge of child protection, regardless of national context.