Representations of child welfare services in Norwegian, Danish and German newspapers

Child welfare services are often scrutinized in media reports, but few comparisons of how services are represented in different countries are available. The aim of this article is to assess whether systematic differences in the representation of child welfare services in Norwegian, Danish and German newspapers can be documented. A content analysis of major newspapers in each country revealed considerable differences among the countries. While Norwegian and German child welfare services are mostly presented as implementing adequately severe and timed interventions, their Danish counterparts are heavily criticized for a complete lack of interventions and for implementing interventions too late or not at all. In none of the three countries, interventions of child welfare services are in the reviewed newspapers presented particularly as too severe or too early. The cooperation with other welfare services is described mostly as neutral or negative in Norwegian newspapers, as positive, neutral or negative in German ones and not at all in most Danish articles. Cultural sensitivity in interventions is in the newspaper coverage of all three countries characterized mostly by an omission of the topic altogether. The comparative indicators applied in this study are suggested as powerful tools for further comparisons.


Introduction
In this article, we investigate how child welfare services interventions that impede private and family life are represented in Norwegian, Danish and German newspapers. Our aim is to assess whether systematic differences can be documented in the representation of child welfare services in newspapers in these three countries.
Cases of grave child abuse and interventions by child welfare services are frequently reported on in the media. In many of these cases, the media coverage of alleged child abuse and maltreatment demonstrates a capacity for creating emotional responses, an awareness of the presence of child abuse in everyday life and an increased sensitivity towards children's rights. Frequently, the role of child welfare services is included in media reports on such cases.
In 2015 and 2016, rallies in several European countries were arranged simultaneously. Protests targeted Norwegian child welfare services taking custody of children from families with immigrant backgrounds, who were allegedly committing child abuse or neglect. In response to these organized protests, the director of the Norwegian Directorate for Children, Youth and Family Affairs described the fundamental dilemma for child welfare authorities as interventions either coming too early or too late (Mjaaland, Fjeld, & Rydland, 2015). This basic dilemma is neither limited to a collision of different cultural understandings of children's rights, nor is it only germane to Norwegian child welfare authorities. The potential for conflicts arising from different cultural perceptions of childhood has been analyzed by Williams and Rogers (2016), among others.
Children's rights, as sanctioned by the General Assembly of the United Nations (1989), emphasize the best interests of the child as the primary consideration for public services in general, and thus, for social welfare services in particular. To fulfill this obligation, child welfare services in most European countries share a common foundation of basic legal obligations (Berg-le Clercq, Bosscher, & Vink, 2012;Schweppe, 2002). Albeit embedded in different national legal frameworks, they share a primary objective in their legal obligation to advocate for and, if necessary, to enforce children's rights when individuals, family or potentially harmful societal structures threaten these rights. This obligation puts child welfare services in latent or active opposition to different stakeholders in children's environments, frequently resulting in conflicting perspectives on what should be regarded as the child's best interest (Križ & Skivenes, 2014).
The terms "child welfare services" and "child protection services" are frequently interchanged. There seems to be no consensus as to what constitutes a clear demarcation between the two terms, even though the former seems to be more commonly used in European countries, while the latter appears more frequently used in North America. As an example, in Norway, no distinction is made between "child welfare" and "child protection services", and the Norwegian term "barnevernet" on government websites translates to "child welfare services" (Ministry of Children and Equality, 2014). Welbourne and Dixon (2013) argue for a distinction not according to the actual service level but to the methodological orientation towards a child protection and a family welfare approach to service delivery. We follow this argument and use the term "child welfare services" for the purpose of our comparison, as this term reflects the methodological orientation in Norway, Denmark and Germany. These three countries represented in our study all share a common orientation towards service delivery. Services are mainly delivered as preventive and supportive services for at-risk families and are provided to those who voluntarily seek out these services, or, in certain cases, are provided as enforced interventions. Furthermore, these countries share an orientation towards family preservation and a simultaneous child-centric orientation in social service delivery (Pösö, Skivenes, & Hestbaek, 2014;Skivenes, 2011;Wolff, Biesel, & Heinitz, 2011). In all three countries, services are reflective of a principle of least intrusive interventions, with a primary aim of solving problems and providing support in-home and in cooperation with families whenever possible. However, out-of-home care rates in these three countries are, in a European context, comparatively high and have continuously increased over the last several decades (Gilbert, 2012).

Media, child welfare services and public opinion
Media representations are an important force in influencing public opinion and policy development and implementation. A considerable body of research has documented the agenda-setting role of the news media (see among others Kunkel, Stacy, Sunding, & Biely, 2006;McCombs, 2004;Rasmussen, Romeijn, & Toshkov, 2018;Soroka, 2002).
According to Asdal (2015), documents create a material-semiotic version of the object they describe. Treating newspaper articles as documents allows individuals to consider the agenda-setting role of the news media as a process of social construction, in which knowledge and hierarchies of value and status are reproduced by the repetition of certain structures of meaning and feeling (Toynbee & Gillespie, 2006 188). Following this line of thought, the prominence or omission given to a specific topic in the media significantly influences the probability of this very topic being reflected in policymaking (Kunkel et al., 2006). Thus, the perspective displayed in the media on interventions by child welfare services following alleged child neglect or abuse can influence public opinion and determine how the public obligation to enforce children's rights is balanced against the right to privacy, family and home. Depending on the focus given to public interventions in the media, these cases of alleged child neglect or abuse can be framed quite differently: as a result of neglect or violation by a child's personal custodian, as a result of ineffective service performance, or as public infringement on individual rights. Thus, media coverage influences the public reputation of child welfare services and the confidence placed in them. Public reputation and legitimacy, on the other hand, influence political legitimacy, which is the basis for the allocation of resources determining working conditions and service performance.

Representations of child welfare services in the media
Compared to the general acknowledgment of the agenda-setting role of the media on the one hand and the attention given to child welfare in the media on the other, the influence of the media on the public's opinion of the performance of child welfare services has received surprisingly little attention. A rather limited number of studies have analyzed media coverage of child welfare in national contexts. Our review of the available studies on the subject revealed the following information.
In an early study, McDevitt (1996) compared, over a 25-year time period, the frequency of media reports on child welfare services in the USA with the frequency of reports of child abuse and neglect to public authorities. Contrary to the results of more recent studies, McDevitt concluded that media reports had not influenced the number of reports to authorities. Instead, the author attributed most of the influence on both types of reports to national policy changes. Ayre (2001) analyzed British newspapers and concluded that they focused mainly on "sensationalist coverage" of extreme cases of child neglect and the misconduct of child protection services. This focus contributed to a general hostility towards child protection workers and to a widening gap between (specialized) child protection services and general child welfare. Due to the potentially fatal consequences in cases of failure, child protection services were given an advantage in the allocation of resources, at the expense of general child welfare agencies.
The study of Niner, Ahmad, and Cuthbert (2013) focused on the media coverage of sexual abuse of children in Malaysia. The media representation was criticized as over-representing particularly horrific cases of child abuse while not addressing the underlying causes of it. The authors themselves attributed these causes mainly to the socioeconomic situations of young and vulnerable mothers. In this study, child welfare services are portrayed as "focused on punishment of perpetrators, while the deeper causes and their appropriate reform are neglected" (Niner et al., 2013 450).
In an Australian context, Lonne and Gillespie (2014) analyzed the print media's role in informing the public about the implementation of a particular government program, laying the groundwork for a strong emphasis on early intervention and prevention policy. The study revealed that police involvement and references to the legal system were mentioned far more often than were child protection services. When child protection services were mentioned, most print media articles implied system failure instead of the misconduct of individual child welfare officials. The findings also indicated a lack of detail in newspaper coverage. The authors argued that this lack of detail, contrary to the intentions of the new government policy, leaves the public poorly informed about the work of child protection services in general and, in particular, about the support provided for families and victims of abuse.
All of these studies investigated the subject within a national perspective. Few comparative studies from a transnational perspective are available. An exception in this regard is a study of Lonne and Parton (2014), who compared the media coverage of child abuse scandals in Australia and England. The study concludes that media representation in England is rather hostile and focused on individual social workers, including managers. In comparison, Australian media representation is described as "relatively tame, generally maintaining a focus on systems rather than individuals" (2014 830). The study illustrates the particular value of transnational comparisons for providing insight into the role of the media in influencing and forming public opinion. However, the focus in Lonne and Parton's work is on the media coverage of particularly grave cases of child abuse and does not reflect media coverage of the work of child welfare services below the threshold of being "scandals".
These studies all document the importance of research on the media coverage of child neglect and abuse and the performance of child welfare services. However, the limited attention that the subject has received indicates a simultaneous and disturbing lack of knowledge about media representation of the interventions of child welfare services in general, and particularly, of cases below the threshold of extreme child neglect and abuse. Furthermore, with the exception of Lonne and Parton's study, systematic cross-national comparisons are not available. In our analysis, we suggest a set of predefined indicators for transnational comparisons that can be used to identify systematic differences in the representation of child welfare services.

Data and methods
We performed a qualitative content analysis of newspaper articles focusing on specific interventions of child welfare services in Norway, Denmark and Germany. The first and second author are both fluent in Norwegian and are native speakers of German and Danish, respectively. This language proficiency allowed us to perform a multilingual in-depth analysis of media reports. The choice of these three countries is thus based on the organizational comparability of child welfare services, as well as our own language proficiency.

Data
We created a database with newspaper articles identified through standardized searches in the digital archives of the following newspapers: Aftenposten, Dagbladet (Norway), Berlingske tidende, Politikken (Denmark), Frankfurter Allgemeine Zeitung and Süddeutsche Zeitung (Germany). All of these are broadsheet newspapers owned by private enterprises, with a national focus and a large national readership. While Aftenposten, Berlingske tidende and Frankfurter Allgemeine Zeitung have a politically conservative orientation, Dagbladet, Politikken and Süddeutsche Zeitung lean towards a social democratic political perspective. Articles were selected from a time window between January 1st, 2014 and December 31st, 2016. A three-year time window ensured that our analysis was resilient to specific cases of child neglect or abuse that, for shorter periods of time, may have dominated the national debate.
A list of relevant search terms was compiled in English. Each term was then translated by the authors into Norwegian, Danish and German. Searches were performed using truncated word stems with wildcards constructed according to the search engines of the newspaper archives. The keywords child welfare service were used in conjunction with the following keywords and the Boolean operator AND: infringement, transgression; violation, abuse; neglect; care, foster, enforce; concern; child* best*, vulnerable. A complete list of the search terms in Norwegian, Danish and German is provided in Table 1.
Altogether, 8372 articles were identified. Since the focus of our analysis was on interventions of child welfare services impeding private and family life, we excluded chronicles, debate articles and articles discussing policy development, legal frameworks and political statements about child welfare services in general and without reference to specific cases. For the same reason, we excluded articles on child welfare services and interventions in institutions, schools and other public arenas, as long as these articles did not also describe specific intervention into the private sphere. We followed the distinction proposed by Hove, Paek, Isaacson, and Cole (2013) between reports on child abuse framed either as isolated, episodic or individual incidents or as a systemic and societal problem, with our focus mainly on the former. However, more thematically framed articles that also mentioned specific cases of intervention were included.
The screening process is described in the following flow chart (Fig. 1).
The 261 articles satisfying our inclusion criteria were analyzed qualitatively. In our analytical strategy, we followed a model for document analysis described by Mayring (2000) as qualitative content analysis and by Kvale and Brinkmann (2009) as meaning categorization. The following analytical steps were performed. We reviewed each of the 261 articles closely and assigned a value to each according to five preselected comparative indicators. Each article was coded according to all indicators. In cases in which different parts of an article indicated contrasting or contradictory coding alternatives, we first coded each part to the appropriate value and then coded the article as a whole to the value most often assigned. When an even number of different coding alternatives was reached for the same indicator, the article was ultimately coded with a value of "unassigned". A value of "not applicable" was assigned in cases in which the text did not yield any information for this particular indicator.
The results of this scoring process allowed us to identify and compare differences in representations of child welfare services in newspaper coverage across the countries. The first and second author are both bilingual proficient in Norwegian. To enhance inter-coder reliability, after agreeing on initial coding criteria, both authors coded the Norwegian articles independently and subsequently compared their completed coding. When discrepancies in coding alternatives appeared,  the authors discussed them until they reached a consensus on more precise coding criteria. Following the joint analysis of the Norwegian articles, the German and Danish articles were coded individually by the first and second authors, who were native speakers of German and Danish, respectively.

Choice of comparative indicators and associated values
We chose five comparative indicators for the purpose of our comparative assessment, each associated with a three or five item scale. The choice of these indicators was determined by two factors. First, we conducted a thorough review of the available literature to the representation of child welfare services in the media. Second, we performed an initial screening of a sample of newspaper articles before we decided on five indicators, which we perceived as suitable to assess systematic differences in the representation of child welfare services in Norwegian, Danish and German newspapers.
The first indicator, "presentation in the media", assessed the overall presentation of child welfare services in the newspaper article. Values assigned were "positive" when service performance was emphasized as particularly good in relation to the case at hand. "Neutral" was assigned when service performance was described as what can or should be expected, and a value of "negative" was assigned to articles criticizing service performance.
We want to present three brief examples in order to allow for a better understanding of our coding strategy. We translated these excerpts from newspaper articles into English and referenced to the original source.
The first example is a sequence which we coded positive.
"To the court hearing at the end of November the mother appeared without attorneythe impartial tone up to now obviously was confidence inspiring. In the hearing, the situation was discussed, a member of child welfare services reported. The parents received an appointment for counselling services already in the beginning of December were they agreed to the modalities of < the child > with his father (Müller-Jentsch, 2015)".
We coded this article as positive due to the description of child welfare services as impartial, confidence-inspiring and swiftly reacting.
The second example is an example for assigning a value of neutral.
«The girls who had been abducted from Moss yesterday morning have been found well in Denmark. The two girls of five and eleven years are in the care of child welfare services. They had been abducted by their biological parents.» (Aftenposten, 2015).
In this case, the intervention of child welfare services is just what has to be expected, since child welfare services have the legal obligation to take care of children whose legal guardians cannot take care themselves. As no information is presented in the article about how in particular child welfare services have fulfilled their legal obligation, we have assigned a value of neutral.
The final example is an article coded to the value of negative on the indicator of "presentation in the media".
"Already in 2005 one of the children told her teacher about violence and threats. Child welfare services were notified in 2006, after two of the children run away from home, telling police and child welfare services about being abused and threatened all their lives. Nevertheless, the case was dismissed. Since then, child welfare services were involved in 2008, 2009 and 2012 after notifications about violence, but all cases were dismissed." (Letvik & Skogstrøm, 2014). In this example, child welfare services are repeatedly presented as quite reluctant to act. In cases like these when child welfare services or their service performance were heavily criticized, we coded the article as negative.
Several research reports have focused on child welfare services acting with overly harsh or inadequate interventions (Gilbert, 2012;Kojan & Lonne, 2012). Therefore, we chose "severity of the intervention" as the second indicator in our analysis. We distinguished between newspaper articles describing a "lack of intervention" and those describing interventions as "not severe enough", "adequate", "too severe" and "inappropriate", based on what the situation (as presented in the article) demanded.
The role of child welfare services in cases of child neglect and abuse, as discussed in the media as well as in the literature, are often related not only to the severity of interventions but also to their timing (Stoltenborgh, Bakermans-Kranenburg, & van IJzendoorn, 2013;Warner, 2013). Therefore, we chose "timing of intervention" as a third indicator and distinguished among articles presenting interventions as being "too early", "adequate" or "too late".
Social and health services are increasingly expected to complement each other in order to provide their service recipients with coordinated and holistic services (Author name excluded for review Schönfelder, 2013;Schönfelder & Nilsen, 2016). Child welfare services, as part of the social service system in the three countries, are no exception in this regard. Therefore, we included the indicator "cooperation with other welfare authorities" in our analysis, with the values "positive", "neutral" and "negative".
In several countries, a lack of cultural sensitivity has been highlighted as a critical barrier in the work of child welfare authorities, particularly with different immigrant groups (Bywaters, Brady, Sparks, & Bos, 2016;Hollekim, Anderssen, & Daniel, 2016). We scored each article on cultural sensitivity using the same values as for the indicator "cooperation with other welfare authorities".
For all indicators, we assigned a value of "unassigned" when the article allowed for different coding alternatives and a value of "not applicable" in cases in which the text did not yield any information for this particular indicator. Table 2 shows the coding frequencies of each of the five comparative indicators we have chosen for our analysis, distributed according to the two newspapers for each country and the chosen timeframe.

Results
For several indicators, the distribution of coding frequencies shows considerable differences between the newspapers of each country. However, Table 2 displays no particular pattern according to the political leanings of the newspapers across the countries. Qualitative analysis revealed that particular cases of child neglect on several occasions had dominated newspaper reports over longer periods of time. Since all the chosen newspapers are distributed nationwide and some also have regional sections, some cases were first reported on a regional level and received national attention later on. Some of these cases appeared continuously during the revised timeframe, referred to as examples of larger tendencies regarding the performance of child welfare services. On several occasions, the work of child welfare authorities was reported on regarding cases that had happened before the timeframe for our analysis. Thus, these cases had an impact on the coding frequencies of the different indicators with associated values.
Since our material is composed of qualitative data, we do not intend to draw statistical conclusions. Instead, our analytical focus is on discussing which inferences can be drawn from the distribution of coding frequencies. To allow data comparisons among the countries, we performed the following methodical steps. First, we merged the coding frequencies for both newspapers in each country. Second, we used a three-year time window to make our analysis more resilient against particular cases that may have dominated newspaper coverage for a certain time. For the same reason, we compiled the relative coding density for each country over the selected time window. Third, we transformed the absolute frequencies to relative frequencies to allow comparisons among child welfare services in the three countries. The distribution of coding frequencies illustrates considerable differences between the three countries for all of the five indicators. In the next sections, we first discuss the specific features of each of the three countries, before we conduct a systematic in-depth comparison of these features.   portrayal of child welfare services in newspapers. The severity of interventions is typically described as adequate, as is the timing. However, newspapers also report on interventions not being severe enough, and in more than one out of four articles, child welfare services are reported as intervening too late. The cooperation with other welfare authorities, such as social services, schools, and health care services, is generally presented neutrally, meaning that the contribution of child welfare services was as it was to be expected, according to the mandate laid out in the legal framework for the services and according to its presentation in the media. At the same time, almost one out of three articles portrays such cooperation as negative. This is a rather disturbingly large number of articles, and this image of the work of child welfare services that is communicated to the public gives cause for concern, as it indicates that the goal of delivering coordinated and holistic welfare services all too often is not achieved. The cultural sensitivity of child welfare workers is typically not remarked upon, indicating that it is not considered relevant in most media representations.

Child welfare services in Norwegian newspapers
In sum, the typical report in Norwegian newspapers portrays child welfare services as neutral and the interventions as adequate in terms of both severity and timing, with no regard to cultural sensitivity. However, child welfare services are, to a substantial degree, criticized for not intervening severely enough and for intervening too late, without cooperating enough with other services. In the data from Denmark, newspaper articles communicate a rather negative perspective on child welfare services. These services are typically criticized for a lack of intervention when needed, with a strong emphasis on untimely delays in cases when action was warranted. Almost one out of five reports present the interventions of child welfare workers as inappropriate for the demands of the situation.

Child welfare services in Danish newspapers
Cooperation with other welfare authorities is rarely mentioned in newspaper reports. This result may indicate that in Danish newspapers, the ideal of coordinated and holistic service delivery does not have a particular impact as a quality indicator of the service performance of child welfare services. At the same time, in one out of five reports, child welfare services are presented as not having cooperated with other services as expected.
Cultural sensitivity is a nonissue in the vast majority of articles, meaning that a typical report does not mention it at all.
Given these results, the typical article reporting on the work of child welfare services in Denmark is quite critical of the severity and timing of interventions, as well as of the cooperation with other welfare authorities, while cultural sensitivity does not have an impact.

Child welfare services in German newspapers
Fig. 5 displays the relative scoring density for Germany. German newspapers portray child welfare services in a more balanced manner than do their Norwegian and Danish counterparts. While neutral or negative connotations dominate, positive presentations are also quite prevalent. The severity of interventions is typically presented as adequate, assessed by what the situation demands, or not severe enough. The timing of interventions is mainly portrayed as adequate as well, even if criticisms for acting too late are also frequent. Regarding cooperation with other welfare authorities, no particular perspective is dominant, as positive, neutral or negative presentations appear with relatively equal frequencies. Cultural aspects are mentioned in some reports but without relating the interventions of services to these aspects. As in the cases of Norway and Denmark, cultural sensitivity as a particular topic is, for the most part, absent, even if about one out of five reports communicate a neutral assessment of the emphasis given to cultural aspects during interventions.
Summarizing the major tendencies in German newspapers, the typical report portrays child welfare services as neutrally or negatively. Interventions are described as mostly adequate but are almost as frequently described as beginning too late and not being severe enough. Cultural sensitivity is typically a nonissue. However, the typical report in German newspapers is far more characterized by variations in the portrayal of child welfare services than is the typical report in Norwegian and Danish newspapers.
In the next section, we discuss the particularities of this distribution for each of the five indicators in more detail, together with the inferences that can be drawn from the findings.

Comparisons of the media representation among the three countries
For all three countries, a critical portrayal of child welfare services newspaper media is far more common than is a positive portrayal. This critical portrayal commonly illustrates a reluctance to act, rather than an overly eager service attitude exceeding the requirements for the severity of the situation or timing of interventions.
Child welfare services are portrayed relatively similarly in Norwegian and German newspapers, mainly as either neutral or negative. In comparison, newspaper reports on Danish child welfare services communicate a much more negative image. Both the similarity between Norwegian and German services and the more critical representation of Danish services are particularly manifested in the indicators of the severity and the timing of interventions. While Norwegian and German child welfare services are mostly presented as implementing adequately severe and timed interventions, their Danish counterparts are heavily criticized for a complete lack of interventions and for implementing interventions too late. In addition, in more than one out of four reports, Danish services are described as implementing inappropriate interventions. However, a considerable number of reports in Norwegian and German newspapers are also rather critical in their accounts of child welfare services for interventions that are insufficient in terms of  severity and speed of taking action. Remarkably, newspaper reports in none of the three countries particularly emphasize interventions of child welfare services as too severe or too early.
The indicator of cooperation with other welfare authorities reveals a more heterogeneous comparative picture. We have argued for the relevance of this indicator above and pointed out that health and social services in all three countries are increasingly expected to cooperate across organizational and professional borders. While the emphasis in Norwegian newspaper coverage is the cooperation between Norway's child welfare services and other welfare authorities as neutral and negative, the German newspaper coverage is more balanced among positive, neutral and negative presentations. Danish newspaper presentations differ considerably, as the overwhelming proportion of reports does not mention inter-organizational or interdisciplinary cooperation at all. In the rare cases in which such cooperation is mentioned, it is displayed as neutral, meaning that cooperation with welfare authorities other than child welfare services happened as expected. We regard the high number of newspaper reports not mentioning the cooperation with other welfare authorities at all as an indicator of relatively little emphasis given to the subject in public discourse.
The last of our comparative indicators, cultural sensitivity, is, in all three countries, characterized mostly by an omission of the topic altogether in newspaper coverage. However, beyond the generally limited attention given to the subject of culturally sensitive service performance, this indicator revealed a somehow less-pronounced difference as well. A noticeable number of reports in the Norwegian and German cases displayed cultural sensitivity as neutral. This finding is somewhat surprising, given the massive protests, particularly against Norwegian child welfare services, mentioned in the introduction of this paper, as these protests took place partly within the timeframe of our analysis. Obviously, the specific cases that caused these protests have not been reflected in the Norwegian newspaper reports on the actual interventions due to a critical lack of cultural sensitivity. In comparison, almost none of the Danish articles yielded any information at all about cultural sensitivity in the performance of child welfare services.

Analytical value of the analytical indicators for further comparisons
The comparative indicators with their associated values documented considerable differences in the portrayal of child welfare services in the newspaper media of the three countries.
The tendencies we delineated from the relative coding density condense and therefore simplify the complexity of the analyzed data. These tendencies should therefore not be confused with the actual occurrence or absence of cases of grave misconduct of child welfare services. During the selected timeframe, Norwegian, Danish and German newspapers have reported on cases of severe child neglect. In the gravest of these cases, child neglect or abuse has resulted in the death of a child, without child welfare services or other authorities acting in an adequate, timely or any manner before a young life most tragically ended. Equally, albeit with a different emphasis, newspapers in each of the three countries had occasionally reported critically on child welfare services with greater severity than the situation required.
The tendencies we delineated from our analysis should be understood as indicative of typical mainstream reports for how child welfare services are portrayed in the newspaper media, rather than as an indicator for the actual occurrence or absence of extreme cases. The results of our study thus provide insight into a variation in emphasis given to different aspects of the work of child welfare services and how their service performance is communicated to the public.
As we pointed out in the beginning of our article, media reports contribute to constructing public discourses about child welfare services. Understood in a Foucauldian sense (Foucault, 2002:131), discourses shape materiality, including the resources allocated to the sector, working conditions for child welfare workers or favored methodical preferences in professional training and practice. In our study, we believe that we have demonstrated the analytical value of the five comparative indicators for framing public discourses on child welfare services in the media. We also believe that we have documented the analytical value of these indicators as powerful tools for further systematic comparisons. The different representations of child welfare services in Norwegian, Danish and German media which we have documented should serve as a point of departure for further in-depth studies that investigate the reasons for these differences.

Strength and weaknesses of the study
The aim of our analysis was to assess whether systematic differences in the representation of child welfare services in Norwegian, Danish and German newspapers can be documented. Our study design provided the means for achieving this aim. At the same time, it limits the possibility for drawing conclusions in two respects. First, our data do not provide insight into the underlying causes for the idiosyncrasies revealed by the analysis in the media's portrayal of child welfare services. Second, our data also do not allow conclusions regarding the reasons for the differences that had become visible.
However, these limitations do not diminish the value of the analytical model as presented. Rather, they indicate the need for further comparative research in which the differences in media presentation, which we have documented, are linked convincingly with an analysis of the reasons for these differences. This does include further analysis of the relation between media presentations as a force for shaping or reflecting policy.
Another area of inquiry to which the use of the analytical model can be extended is the study of other fields of public service delivery, in order to assess whether the tendencies we have documented are reflected, particularly in other sectors of social and health services delivery.

Conflict of interest
None.

Funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Ethical considerations
The study is based on publicly available newspaper articles and did not involve human subjects (including human material or human data). In compliance with the Helsinki Declaration, an approval of an ethics committee was not required.