Time, Institutional Support, and Quality of Decision Making in Child Protection: A Cross-Country Analysis

ABSTRACT This paper examines perceptions of time and institutional support for decision making and staff confidence in the ultimate decisions made—examining differences and similarities between and within the service-oriented Nordic countries (represented by Norway and Finland) and the risk-oriented Anglo-American countries (represented by England and California). The study identifies a high degree of work pressure across all the countries, lines of predominantly vertical institutional support and relatively high confidence in decisions. Finland stands out with higher perceived work pressure and with a horizontal support line, whereas England stands out with workers having a lower degree of confidence in their own and others’ decisions.

insufficiently prepared to engage in sound decision making (O´Sullivan 2011). Relatedly, large caseloads, reported in many European countries and an endemic problem in many US states, have been linked with poor child welfare outcomes (United States Department of Health and Human Services, 2004;Wagner, Johnson, & Healy, 2009); in particular, concerns have been raised that large caseloads reduce staff time with individual families, thereby compromising the quality of important child welfare decisions (US GAO, 2003). In these highly consequential circumstances, too little time may compromise the amount and adequacy of information available, though how much time is sufficient for information gathering has not been revealed in the literature. In fact, little information is available in the published literature at the descriptive level to characterize the amount of time available for decision making in most of the states represented here.
In this paper we examine the perceptions of child protection workers in the United Kingdom (England), Finland, Norway, and the United States (California) regarding the time available for deliberation, institutional support, and confidence in the quality of decision making relating to child removal. We are unable to assess the quality of the ultimate decisions made, in part because the outcome of interest (e.g., child maltreatment versus compromised child well-being) varies greatly across our four-country context; the goal instead is to pursue questions that shed a brighter light on the comparative welfare states in which child protection systems are nested and to learn from the similarities and differences revealed in the findings.
The paper is based on data from an online survey of 772 child protection workers experienced in preparing care order preparations as part of their work at a child protection agency. Gilbert and associates have elsewhere outlined the similarities and differences between these countries (2011) and have characterized Finland and Norway broadly as family service oriented and England and the United States as child protection oriented, though these formerly stark distinctions (Gilbert, 1997) may be blending over time. England, in particular, seems to be something of a hybrid, with family support services delivered within a child protection focus. We use the term child protection workers across the countries, even though each may use different terms to describe frontline staff in their child welfare system. We refer to care order preparations to indicate the staff practices related to decision making that result in an application to court for an involuntary removal. The policy framework shaping these four state systems for decisions in care order cases are characterized elsewhere (Berrick, Peckover, Pösö, & Skivenes, 2015).
In the following section we present other research relating to time, institutional support, and decision-making quality followed by a presentation of the four system contexts and the hypotheses we derive about the three research themes. Methods and Findings sections come next, and the paper ends with a discussion and a brief conclusion.

Time, institutional support, and quality in bureaucratic contexts
Legislation that defines the threshold for involuntary intervention shapes how and when child protection workers interact with a child and his or her family (e.g., Benbenishty, Osmo, & Gold, 2003;Kriz & Skivenes, 2013, 2014Skivenes & Stenberg, 2013). The legal system also has an impact on child protection workers' decision making and thresholds for care orders and sets a frame to which workers must be responsive (Dickens, 2006;Hawkins, 2002). Regardless of the legal requirements for a response, child protection staff have long raised concerns about the pressures on the time they have available for decision making and the consequences of over-and underinvolvement (e.g., in England, Munro, 2011). More recent literature has also highlighted the importance of timely decision making from the perspective of the child. United States and English scholars in particular (e.g., Bartholet, 1999;Berrick, Barth, Needell, & Jonson-Reid, 1998;Brown & Ward, 2013) have urged earlier discovery of serious child maltreatment and more swift and decisive action to respond to child and family concerns. Child protection staff have articulated the negative consequences that can befall children when delays in decision making compromise children's health or well-being (e.g., Beckett, McKeigue, & Taylor, 2007).
In time-pressed agency environments, research shows that staff often develop strategies for managing excessive demands (Smith & Donovan, 2003;Munro, 1998Munro, , 2011; whether these strategies result in compromised decisions for families may be a concern. Certainly given the complex and unique characteristics that many child welfare service users bring to practice, the challenge of ensuring equitable and consistent decisions between and across cases may be especially difficult. In some jurisdictions, an emphasis on team decision making is used as an institutional strategy to ensure greater uniformity of decisions across cases; in other settings, assessment tools are used increasingly to help shape uniform decisions (Crea, 2010); elsewhere, some combination of teambased and standardized measures may be employed. In short, in time-constrained environments, child protection staff need institutional support for reasoned decision making. Such support may take the form of organizational structures and/or emotional scaffolding upon which staff can rely (Miller & Fisher, 1992) and be held accountable.
Accountability is a central concern in child protection in part because of the gravity and consequences of these public actors' decisions; the power interplay between child protection staff and the typically marginalized populations served by child welfare; and the contested nature of interventions into the private sphere. Accountability is commonly used as a standard for good governance (Bovens, 2007) and includes concepts such as transparency, legitimacy, trustworthiness, responsiveness, and justice (Bovens, 2007;Mulgan, 2000). According to Bovens (2007), accountability is "a relationship between an actor and a forum, in which the actor has an obligation to explain and to justify his or her conduct, the forum can pose questions and pass judgement, and the actor may face consequences" (p. 450). Actors may be individuals or organizations. In modern democracies there are typically multiple, institutionalized accountability mechanisms. In an analysis of the policy structures guiding decision making in child protection in the four countries under study, Berrick et al. (2015) identified the role of the public-administrative and professional accountability domains. At the system and organizational levels the authors identified an auditing approach to monitor the quality of care orders (e.g., California, Norway, England) and at the individual level a legislated emphasis on service users', social work colleagues', and managers' control (e.g., Finland, Norway and England). Depending on the system context, the structure of accountability mechanisms and institutional support can be vertical or horizontal (Lindberg, 2013).

Four country contexts for decision making
Child protection workers' experiences as decision makers regarding care order preparations are shaped by the institutional framework in which they are employed. Case-workers' decisions on individual children are influenced by macro-elements such as the legal, welfare state and child welfare systems (Benbenishty et al., 2015;Duffy & Collins, 2010;Smith & Donovan, 2003); this intersection is the focus of this study. In a family service oriented system (e.g., Finland and Norway and, to a degree, England), child protection workers typically have long-standing relationships working with families that may last over months or years. Their goal, to promote child well-being, allows them and other multiprofessional staff to track families' progress over time; decision making relating to care orders may therefore unfold gradually. Finland relies on a negotiation process in collaboration with coworkers, children, and parents to determine an appropriate decision; Norway relies on team leaders and team managers to review decision making (Berrick, Dickens, Pösö, & Skivenes, in press). In a child protection system such as the United States (more specifically, California, the site for this study) and England (dominantly), decision making is more heavily weighted toward notions of risk and safety. High thresholds for state involvement may dictate urgency in decision making and vertical processes for confirming decisions. These rights-oriented systems ensure that all parties have legal representation; in England, layers of multiple procedures are mandated to ensure quality control across cases, and, as far as possible, to divert cases from court action; in California, use of evidence-informed risk-assessment tools are employed for the same purpose (Berrick et al., in press). We expect findings from this study to align such that workers in Finland and Norway will offer somewhat similar responses regarding time, support, and assessments of quality decision making, and their perspectives will differ from workers in England and California, whose responses will also broadly align.
In order to examine how child protection staff interpret and experience the formal frameworks within which they work, we examined three important dimensions of decision making including time, institutional support, and perceived quality. Given the child welfare systems within which our respondents were situated, we anticipated that staff from California would feel the press of time most acutely compared to their international peers. California and England, as relatively reactive systems with high thresholds for assessing risk and harm, were expected to impose more extreme time demands and to use vertical structures of support and accountability to authorize decisions. Predicting workers' likely perceptions of the quality of their work was not a straightforward process. Highly proceduralized systems, such as England's, or systems with standardized-decision-making tools, such as California's, could raise confidence (decision making based on routinized processes) or weaken it (decision making privileging processes over professional skill and knowledge). Workers' responses could be influenced by many factors, including organizational, professional, and societal support for social workers who engage in child protection. In England, where such social support tends to oscillate, often depending on the latest public scandal (Butler & Drakeford, 2005), it has long been recognized that the proliferation of procedures does not guarantee immunity for staff if things go wrong (Howe, 1992;Munro, 2011;Parton, 2014).
Finnish and Norwegian responses were expected to align and to show staff perceptions of sufficient time for decision making and ample institutional support due to the length of the processes, the nature of their service-oriented systems, and their team-based approaches. We anticipated that the authorization for care order applications would be relatively horizontal compared to England and California, and thought it likely that staff from Finland and Norway would feel confidence in the quality of their and their coworkers' decisions due to a high degree of professional discretion and the various institutional supports available for their decisions.

Methods
Findings presented here come from a larger study funded by the Norwegian Research Council relating to decision making in child protection in England, Finland, Norway, and California. The study includes a sample of 772 child protection workers with experience with care order processes who completed an online survey including questions about time, accountability, and the quality dimensions of decisions. Child welfare workers can be challenging to recruit for study due to considerable work pressures and high caseloads. In addition, it is not possible to get access to child welfare workers with statutory duties in a similar way in the studied countries due to the different ways of organizing and providing child welfare services. As such, we developed a recruitment approach customized to each country context. All subjects were contacted via email and asked to complete an online survey accessed via an electronic link. In Norway, researchers were given email access to the worker union ("Felles-organisasjonen" (FO)) for all child protection members (approximately 1,500), yielding a response rate of 30%. In Finland, the trade union for professionals working in social welfare (Talentia) distributed the email. Trade union membership is high in Finland, however, the association does not categorize membership by type of employment. It is estimated that there are about 2,700 social workers in the Finnish municipalities; only some of them work in child welfare (Mänttäri-van der Kuip, 2015). However, their number is not exactly known. As such, we cannot state the number of social workers employed in child welfare who might have been invited to participate and therefore cannot calculate a response rate. In England, the survey was initially distributed via two representative bodies for social workers, the British Association of Social Workers (BASW) and the College of Social Work (TCSW). These are not trade unions, and social workers are not required to be members of either. Both organizations also include social workers from adult care services. Due to a limited response from this strategy, an alternative recruitment strategy via snowball sampling was employed. Social workers on the contact list of the School of Social Work at the University East Anglia were contacted and asked to complete the questionnaire and/or forward the message to colleagues who might be interested. There was the offer of a £10 shopping voucher for the first 50 to complete the questionnaire. We are unable to calculate a response rate as the number of potential study subjects who were invited to participate is unknown. In California, 10 Bay Area counties participated in the study. All emergency response and dependency investigations social work staff were sent an email from researchers, distributed by their agency manager. The email included the study invitation and a link to the online survey (n = 260). Respondents were offered a US$20 grocery gift card. Ninetyeight (38%) social workers responded.
Of the total 1,020 informants who responded to the survey, 772 had experience with care order preparations and thus were eligible for the study. The total numbers of respondents from each country were 367 from Norway, 208 from Finland, 102 from England, and 84 from the United States. Further details about the data material and the process can be obtained at the following web address: http://www.uib.no/admorg/85747/survey-material#.
Respondents had worked in the field of child protection for an average of 5-9 years, though the English respondents had worked in it for the shortest average duration at 1-4 years. Respondents from Norway and England were, on average, younger (both 31-40) than their peers in Finland and California (both 41-50). The large majority of respondents from California (91%) possessed a master's degree in social work. In Finland, 65% had an MA degree (suggesting that upwards of 35% of respondents in Finland were unqualified temporary workers since they lacked the MA degree) 1 ; in England, 57% had an MA degree; and in Norway, 9% had an MA degree. The large majority of respondents across all four countries were female, though in England, 36% of respondents were male. Among all children's services staff in England, 15% are male (Department for Education, 2013), so the gender distribution in this survey is notably different from the country average.
The online survey was answered between February 2014 and June 2014. The survey took approximately 8-12 minutes to answer. The project was peer reviewed as part of the application process for funding from the Norwegian Research Council, and ethical approvals were confirmed in each country according to national or university-based requirements. The survey questions were first developed in British English by the four researchers so that they were relevant in each child welfare system. The questions were then translated into Finnish, Norwegian, and U.S. terms. The translations into Norwegian and Finnish were controlled by an independent source. The survey was tested by a small group of social workers in each country to ascertain that the questions and the vignettes were realistic in each country.
Study participants were presented with the following case vignette: You are working with a boy-Alex-who is 5 years old and whose family has received in-home services over a period of time. The case includes parental substance abuse, previous domestic violence, and general neglect. The circumstances of the case have deteriorated recently to such an extent that you are concerned that the boy's risk of harm is high. You are starting preparations for care order proceedings with a view to removing Alex from his parents, and you have an interview with the parents to inform them about this. The parents are opposing a removal of Alex.
The survey included questions pertaining to time: "Given the circumstances at your workplace, would you have sufficient time in the case of Alex to undertake all the preparations for court to your satisfaction?" Answer categories were yes (coded 1), no (coded 2), and I don´t know (coded missing). Throughout the analysis we examine whether there are systematic differences between the workers who perceive they have sufficient time versus those who do not. We also asked: "Approximately how long would you spend on the following processes (based on your experience with similar cases)." The 1 We are aware that "temporary" and "working in the system for 5-7 years, on average," are contradictory. However, in the Finnish system one can work as a social worker without the proper qualifications for 2 years within one municipality and then go to another municipality and work for an additional 2 years, and so on. The word unqualified thus has a certain meaning and the word temporary, another meaning, both relevant for the Finnish context. answer alternatives were (1) from the time you informed the parents that it may be necessary to remove the child until a decision is made to remove him and (2) from the time the decision to remove is made until the court report is due in court. An ordinal scale is presented with values ranging from 24 hours to 61+ weeks. The value "can´t say" is coded as missing, and the remaining data are treated as ordinal variables with the following coding: 24 hours (1), 48 hours (2), 72 hours (3), 96 hours (4), 5 days (5), 6 days (6), 1 week (7), 2 weeks (8), 3 weeks (9), 4-6 weeks (10), 7-9 weeks (11) continuing up to 58-60 weeks (28), 61+ weeks (29).
The survey also included two questions pertaining to institutional support and accountability: "Who authorizes your decision to remove a child?" and "Who thoroughly assesses the quality of your work and decision making in child removal?" The answer alternatives were "no one," "coworkers," "supervisor," "senior manager," "agency lawyer," "independent source," "interdisciplinary team," "parents of the child," "child," and "other." Respondents could select all that applied. Coworkers, interdisciplinary team, parents, and child were considered horizontal sources of institutional support; supervisor, senior manager, and agency lawyer were deemed vertical.
There were three questions pertaining to the worker's perceived quality of decisions made: "Based on your experience in the last 12 months, how confident do you feel in the quality of (1) the decisions you make . . ., (2) the decisions your colleagues make . . ., and (3) the decisions your managers make on cases concerning court ordered removals?" Answer alternatives were given on a Likert scale, from very skeptical, coded as 1, to highly confident, coded as 5.
We also examined the impact of caseload, assessed with the question: "What is your caseload today, measured by number of children?" Caseload, measured by number of children listed in a range, was converted into an ordinal variable: 1-3 children coded as 1, 4-6 children coded as 2, 7-9 children coded as 3, 10-12 children coded as 4, and so on to 99+ children coded as 34. The gender variable was coded into a dichotomous variable with the values female (0) and male (1). For these analyses the education variable was recoded from a set of dummy variables into a categorical variable consisting of three values; BA (1), MA (2), PhD (3), where the value "Other" was coded as missing. Work experience was measured narrowly as the respondent's experience as a child welfare worker, asking, How many years have you worked as a child welfare worker, and was coded as follows: Less than 1 year (1), 1-4 years (2), 5-9 years (3), 10-14 years (4) and 15+ years (5).
We used SPSS to examine descriptive statistics, correlations, and mean comparison t tests. We also used the data tool Zigne to assess levels of significance, based on a cluster sample and use of a two-sided test. In the paper we report on significant differences of p < .01 (= ***) and p < .05 (= **), well aware that p < .05 is on the margin of what is relevant to report. An appendix that is posted online as supplementary material provides the various analyses in detail 2 .

Time for decision making
Did workers indicate they would have sufficient time to prepare Alex's case to their satisfaction? Compared to staff in the other countries, a significantly smaller proportion (p < .01) of Finnish staff indicated that they would have sufficient time to prepare the described care order case (see Table 1. Given the circumstances at your workplace, would you have sufficient time in the case of Alex to undertake all the preparations for court to your satisfaction? Responses by country, Percentage and n. N = 756.

Finland
Norway   Table A1 in Appendix). Less than one-quarter of child protection workers in Finland thought that they would have enough time. Somewhat more than one-third of Norwegian workers felt as though they would have enough time, given the context of their workplace; about two-fifths of California workers indicated that time would be sufficient in their work context, and about onehalf of English staff felt the same. How much time would workers spend on preparing that case? Although a smaller proportion of Finns and Norwegians indicated that they would have enough time to prepare this case for care order proceedings, they nevertheless appeared to spend more time on the work than staff in England and California (see Table 2). Examining the median values we find that half of the workers in England and California anticipated they would spend 48 hours or less (California) or 96 hours or less (England) from the time they told the parents that they were starting preparations for a care order application to the time the final decision was made. Workers in Finland and Norway would spend much more time; more than half of the staff anticipated they would spend 3 weeks (Finland) and 4-6 weeks (Norway). These differences between the Nordic countries and California and England were statistically significant (p < .01) (cf. Table A2 in Appendix). 3 Similar statistically significant findings are evident for the time between determining that a care order application was necessary and when they would file their application in court (p < .01) (see Table 2 cf. Table A3 in Appendix). About half of the Finnish and Norwegian workers indicated that they would spend between 4 and 6 weeks on this process, whereas about half the staff in England and California would spend 5 days or less.
Examining the relationship between workers indicating whether they do or do not have sufficient time to prepare the Alex case to their satisfaction, we find that, in general, those who spent less time on the case were more likely to indicate that they had sufficient time for decision making. This approached significance for the Norwegians (p < .10) and for the Finns (p < .05). Tables A4 and A5, provided in the appendix, display mean and median, comparisons, and two-tail t tests. Although the exact meaning of a "caseload" in child welfare is a contested topic (Child Welfare Information Gateway, 2010; Yamatani, Engel, & Spjeldnes, 2009), we framed the question for staff using the wording "your caseload today, measured by number of children." We find that the Finns had the highest caseloads, with a median of 46-48 children; whereas staff in the other countries had caseloads less than half the size (see Table 3). Analyzing the relationship between workers' caseload and their perceptions of the sufficiency of time, we find one significant difference: Finns who indicated they had sufficient time had fewer children for whom they were responsible (p < .01). (Table B1 and B2, provided in the appendix, shows mean and median responses, mean comparisons, and two-tail t tests.)

Institutional support for decision making
Child protection staff do not make decisions without the authorization of others within or associated with their agency. When asked, "Who authorizes your decision to pursue a care order?" all of the respondents indicated that someone else was involved (see Table 4). Supervisors or senior managers were noted by the large majority of all respondents (California, 88%, and Norway, 90%), though English and Finnish staff were somewhat less likely to indicate that supervisors or senior managers authorized these decisions (70% and 69%, respectively). Beyond supervisors and managers, others were also noted as involved in these authorizations. Finland's staff were more likely than staff in other countries to indicate that their peers and/or family members (i.e., the child or the parent(s)) would be involved in authorizing their decisions. About one-quarter of Finnish staff (28%) indicated that coworkers would be involved; only about 5% of respondents in Norway and England and virtually none of the California staff indicated that coworkers would have such authority. The Finns were also more likely to evidence their teambased decision-making processes, as 19% indicated that multi/interdisciplinary teams would be involved in authorizing the decision. This was also the case among about one-fifth of the California staff (22%). And finally, the legalistic approach of the English and California systems were evident as respondents noted the involvement of attorneys in authorizing decisions to remove: 40% of respondents in England and 19% in California noted that lawyers authorized their decisions to go to court to request a care order. When asked, "Who would assess the quality of your decision?" the large majority of respondents in California (92%) indicated that a supervisor or senior manager would assess their decision. These responses were less frequent among English (54%), Finnish (60%), and Norwegian (65%) staff. Again signaling the legal framework of the Norwegian, English, and California systems, lawyers were noted as playing a role assessing the quality of child protection staff's decisions (29%, 42% and 23%, respectively). In Finland and Norway, respondents were much more likely to indicate that coworkers played a role assessing the quality of their decisions (47% and 25%, respectively).

Confidence in care order decisions
In addition to the context of decision making, we were eager to learn whether workers were confident in the decisions they and other agency actors made pertaining to care orders (see Table 5, cf. Appendix Tables C1, C2, C3 and C4). Staff in Norway, California, and Finland were confident or highly confident in the quality of their own decisions. English staff stood out with significantly (p < .01) different responses. Norway and Finland maintained their high confidence in their colleagues' decision making. California staff were significantly (p < .01) less likely than the Nordic workers to hold such high regard for their coworkers' decisions, and English staff were significantly less likely than all their Nordic peers (p < .01) and Californian peers (p < .05) to show confidence in their colleagues. The same pattern held when asked about their managers' decisions: Norwegian responses stand out as statistically significantly different from Finland (p < .05) and the English and California workers (p < .01), with respondents showing confidence in their managers' decisions. Finnish staff were also fairly confident in their managers, whereas California staff were significantly less confident (p < .05) than their Finnish peers. Finally, English staff continued to show skepticism and had significantly (p < .01) less confidence than all the others in the decisions of their managers.
Examining whether any of the background characteristics of workers such as gender, education, work experience, or caseload are correlated with workers' confidence, we see few significant relationships (details are provided in the Appendix, Tables D1, D2, D3, and D4). However, gender, work experience, caseload, and education matter in some countries. Female English staff were more likely to express confidence in their own decisions (p < .01). Work experience was of importance in the samples of English and California staff. The number of years of experience corresponded to a higher degree of confidence in the worker's own decisions (p < .01). English staff with less education were more confident in their own (p < .01) and their managers' decisions (p < .01), whereas Norwegian staff with higher education had more confidence in their own decisions (p < .05). Caseload is significant for English (p < .05) staff´s confidence in their colleagues' decisions, as those with higher caseloads show less confidence in their colleagues' decisions among the English staff. For the Californian staff, higher caseloads are correlated with a higher degree of confidence in their colleagues' decisions.

Limitations
Limitations to the study involve, most obviously, that a survey of this nature only measures what workers say they do or would do, rather than what they actually do. Furthermore, it was especially challenging to devise questions and find terms that were meaningful across all four countries; and as with all questionnaires it may be that respondents interpreted the questions in ways that we did not anticipate. The sample is relatively large and includes child welfare workers across different countries, agencies, and local contexts. In this regard, the study approach, questions, and findings are unique. We acknowledge, however, that the response rate was either smaller than ideal or unknown in the case of Finland and England. Findings may not be representative of all social workers' views in the selected sites. Nevertheless, the findings can be set in the context of other studies about child welfare practice in the four countries, and this provides grounds for drawing tentative conclusions regarding policy and practice experiences across country contexts.

Discussion
Child protection workers in California, England, Finland, and Norway make decisions regarding care orders in very different systems and contexts, and it can be expected that these systems will shape the amount of time allowed for decision making. Perhaps not surprisingly, when presented with the same case scenario, about one out of two workers state that they do not have sufficient time to prepare the care order application to their satisfaction. An analysis of obstacles for making and justifying good decisions shows that time pressure is perceived as a significant obstacle. Workers allege that time pressure negatively influences the quality of the court application; other studies show that time limitations may impact staff interactions with the child and his/her parents (Juhasz & Skivenes, in press). The data suggest markedly different time frames for decisions. In California, regulatory time frames for decision making are very tight: within days of determining that a removal is necessary, staff bring their recommendation to court (Berrick et al., 2015). In England, there is no specified time scale, but there is considerable policy pressure to reach decisions promptly in "edge of care" cases. In Norway, the deadline is 3 months, and almost all Norwegian workers reported that they would conduct their work within this time frame, although with clear variation between workers. And in Finland, there are no regulatory time frames dictating this aspect of care order preparations (Berrick et al., 2015). In spite of these differences, we found respondents in Finland most likely to feel the press of time, followed by the Norwegians. The Finns reported time frames of 4-6 weeks (median) to prepare a court application, yet they were the most likely to indicate that time was not sufficient to undertake all the preparations for court to their satisfaction. The Finns also reported the highest caseloads; within Finland, staff with the highest caseloads were more likely to report that they would not have sufficient time, an important indicator of work pressure (cf. Cole, Panchanadeswaran, & Daining, 2004;Steen, 2010). The variability between staff in the four countries may suggest that workers are engaged in different activities during this deliberative process (as is the case in Finland); that different factors are necessary to consider due to differences set by legislation; and that different thresholds of risk or concern may be at play. Some studies indicate that workers in different countries identify many of the same risk factors in a case but that the assessment of the level of risk and the actions they take depends on the legal and service options and processes available (Skivenes & Skramstad, 2013;Skivenes & Stenberg, 2013). It is also a finding that the managerial role is important for organizing work pressure (Wilson, 2009, cf. Juhasz & Skivenes, in press).
Decisions regarding care orders are weighty; in all of these countries staff made these decisions with the support or supervision of others. In California, England, and Norway, we would characterize the institutional support for decision making as vertical; staff were largely reliant on supervisors, managers, or lawyers to approve their decisions. The rights-based systems of these countries sometimes require layers of multiple authorizations. We would characterize the Finnish process as horizontal, where staff were likely to indicate a wider range of individuals involved in decision making, including coworkers, interdisciplinary/multiprofessional teams, parents, and/or children.
Where staff are asked to rely on multiple actors to review and consider the circumstances of a case (such as Finland), we would expect workers' confidence in the quality of decisions to be high. Similarly, where staff use evidence-based tools to help assess the likelihood of risk (California), we would also anticipate a high degree of confidence in the decisions that are made. Findings from this study were rather contradictory in this regard. Staff in Norway conveyed a very high degree of confidence in theirs and their colleagues' decisions; this, in a system where high professional discretion prevails. In California, staff showed confidence in their own decisions (as expected), but were less sanguine about the skills of their colleagues. Use of evidence-based tools help to confirm individual professional judgment, but if staff do not have confidence in the uniform utilization of these tools by other staff, their confidence in others may be less robust. England stands out, with its staff showing the least confidence in the quality of their own and others' decisions. Perhaps one reason for this is that among the four countries, England has the most tightly regulated and highly proceduralized child protection system, which has been criticized for impacting staff confidence and morale (e.g., Baginsky et al., 2010;Munro, 2011;Parton & Berridge, 2011). Public intolerance for errors made in the decision-making process has also been examined as having a negative impact on social workers in England (Parton, 2014).

Conclusion
We were interested in determining whether there is coherence between the child welfare state orientations of family services and child protection that might be evident in frontline child welfare practice. In general, the findings support some of our initial hypotheses based upon the wellestablished child welfare frameworks within which staff conduct their work. The welfare state in which frontline workers are situated exerts influence through the formal, legal frameworks that shape practice. So too, we see these child welfare system contexts molding frontline experiences and perceptions. In some countries, these connections are relatively clear. In a child protection system (such as California), where decisions are based clearly on imminent risk of harm, determinations about child safety can and should be made expeditiously. Because of the pressures of time in an imminent risk context, vertical structures for assessing and authorizing decisions may be appropriate; time for deliberative team-based decisions may place children at even greater risk of harm. And where decision making is guided by evidence-based tools, one might expect high confidence in individual decision making.
In family service systems that attend to children's well-being (such as Norway and Finland), determining when the threshold of poor well-being has been crossed may be elusive and may well change over time. "Well-being" is a normative construct (some might argue much more than child "safety") that is shaped by culturally agreed upon standards. As such, reliance on horizontal structures to assess and authorize decisions may be an appropriate, though time-consuming, strategy for verifying these norms. The Finnish data appear to comport with such a theory; findings from Norway, however, are not as easily explained.
Where England falls along this continuum might be debated, which may be related to the highly politicized context of the English child protection system. Central government policy and guidance is ambiguous, on the one hand encouraging a broad approach to serving children in need and on the other pressing for narrow interpretations of safety following high profile child deaths (Parton, 2014). These forces position England and its staff at times (often simultaneously) in a child protection system oriented toward risk and safety, and in a family service orientation more closely aligned with, or at least aspiring to be aligned with, the Nordic countries (Parton & Berridge, 2011).
What emerge from the data are aspects of child welfare practice that reflect the larger systems in which staff are embedded. The international comparative value arises as we consider whether approaches of other country contexts might be appropriately imported from one state to another-whether, for example, a horizontal frame for decision making would serve the safety needs of children in California or whether a vertical frame would serve the normative calibration requirements of Finland. As we examine opportunities for transfer of ideas and approaches, child welfare states across the globe are undergoing transformation (Gilbert et al., 2011). As states experience the globalization of child welfare, the export of ideas will need to be tested within the national frame of their new context. Findings from this study suggest that national systems and contexts are still highly influential on the perceptions and experiences of frontline staff. As the philosophical and policy landscapes continue to evolve, child protection staff will have to respond appropriately to the new requirements of practice. The findings offer insights into some of the ways that child protection workers in four countries experience their work in its diverse organizational and policy contexts and suggest profitable areas for further international and comparative research.

Funding
The research project was funded by the Norwegian Research Council. Table A2. Two-tailed independent samples t test of significant differences in countries' mean responses to A: Time spent, from informing the parents of the possibility until decision is made. Reporting p values and level of significance: p < .01 = ***, p < .05 = **, and p < 0.1 = *.

Finland & England
England & Norway

England & California
Norway & California Significant difference in mean response 0.023** 0.000*** 0.000*** 0.000*** 0.000*** 0.000***  Table A1. Two-tailed independent samples t test of significant differences in countries' mean responses (yes/no) to question: Given the circumstances at your workplace, would you have sufficient time in the case of Alex to undertake all the preparations for court to your satisfaction? Reporting p values and level of significance: p < .01 = ***, p < .05 = ** and p < 0.1 = *.

Finland and Norway
Finland and California

Finland and England
England and Norway
0.000*** 0.037** 0.032** 0.000*** 0.000*** 0.000*** Table B2. Two-tailed independent samples t test of significant differences between means in "yes" and "no" group for question about sufficient time for mean values on question: What is your caseload today, measured by number of children? Reporting p values and level of significance: p < .01 = ***, p < .05 = **, and p < 0.1 = *.  Table C4. Two-tailed independent samples t test of significant differences between means in "yes" and "no" group for question about sufficient time categorized on questions considering confidence in decision making per country. Reporting p values and level of significance: p < .01 = ***, p < .05 = **, and p < 0.1 = *.
The decisions you make on cases about seeking care orders in court The decisions your colleagues make on cases about seeking care orders in court The decisions your managers make on seeking care orders in court  Table D1. Correlation between background variables and confidence in the decisions you (A), your colleagues (B) or your manager (C) make on cases about seeking care orders in court using Kendall's Tau B and Kendall's Tau C correlation tests. Reporting Kendall's Tau B and C correlation coefficients and level of significance: p < .01 = ***, p < .05 = **, and p < 0.1 = *.   Table D4. Independent samples t test of significant differences in gender mean responses to questions on confidence about the decisions you (A), your colleagues (B) or your manager (C) make on cases about seeking care orders in court? Reporting p values and level of significance: p < .01 = ***, p < .05 = **, and p < 0.1 = *.