In the Periphery: Ethical Considerations When Indirectly Involving Children in Research

ABSTRACT

In the last decade, there has been an increased focus on child protection cases in the European Court of Human Rights. The heightened attention on child protection underscores the actuality of children’s rights, rendering it an interesting area for researchers in fields such as children’s rights, child protection, or international relations. All judgments from the Court are public. Research on social media data has shown that public data is not necessarily intended for the public. Moreover, in child protection cases, the child is rarely a part of the case. It is the biological parents against the domestic authorities. Nevertheless, the situation the case revolves around includes the child, making the child an indirect part of the judgment. There is little research on the ethical considerations of indirectly involving children in research on court judgments. However, there is abundant research on ethical considerations in researching public data, specifically social media data. By drawing on research on ethical considerations in the use of social media data, this article delves into the ethical consequences of including children indirectly in research. The article pinpoints crucial areas that require special attention from researchers and provides insights on navigating ethical challenges effectively.

KEYWORDS:

• Child protection
• indirect involvment
• children
• child welfare
• court judgments

Introduction

In the last decade, there has been an increased focus on child protection cases in the European Court of Human Rights (ECtHR). When a case is brought before the Court, it is – in most cases – the biological parents who are the applicants. Consequently, children may become unwittingly or perhaps unwillingly involved in the case through the representation of their parents. Moreover, the heightened attention on child protection by the ECtHR underscores the significance of children’s rights, rendering it an intriguing area for legal researchers, social scientists, and others interested in fields such as children’s rights, child protection, or international relations. However, ethical hurdles must be overcome regarding the indirect involvement of children in research, as the increased focus on child protection cases may lead to increased attention on children in vulnerable positions.

The National Committee for Research Ethics in the Social Sciences and the Humanities (NESH) in Norway has created advisory guidelines which aim to contribute to ‘developing ethical judgement and reflection, clarifying ethical dilemmas, promoting responsible research, and preventing misconduct.’ (NESH, 2021). The guidelines underline that ‘Children who participate in research have a particular right to protection’(NESH, 2021, 17). Moreover, research on vulnerable groups, cultural differences and in areas of high tension (which all may play a role in research tied to CPS) demands heightened ethical scrutiny (NESH, 2021, 31 and 32; cf. Ringdal 2013). The NESH guidelines also emphasise that researchers should, in most cases, get consent from those involved in the research, respect privacy and family life, ensure anonymity (if prudent), and underline that the researcher has a responsibility towards persons who are directly or indirectly affected by the research (NESH, 2021, 15, 20, 23, and 26, see also 16). There are exceptions, however. If the data is publicly available or obtaining consent is ‘impracticable or impossible’, the requirement for consent is dropped (NESH, 2021, 18). If so, the researcher is obliged to balance the ethical implications of not having consent against, for example, the negative influence on the verifiability of the research (Brinkmann and Kvale 2017). As the NESH guidelines illustrate, research involving children and CPS poses a multitude of ethical hurdles for researchers to navigate.

A parallel can be drawn to another research area where similar ethical hurdles can be found: social media. Twitter, Facebook, Instagram, TikTok, and so on created places where the average Joe or Jane can make their unfiltered opinions public. However, statements made publicly are not synonymous with statements intended for the public. When researching court judgments or social media, the researcher is faced with analysing data that the respondent is most likely unaware can be used for research. The statements may be private, but the data itself is public, making consent not required. Moreover, even though the data is deidentified through an online screen name, there is still a theoretical possibility to reidentify the data. Thus, the researcher must consider the ethical implications of their research against the risk of harm for those involved – both directly and indirectly. In short, the researcher must consider whether to protect individuals and those connected with the individual from themselves (cf. Barrett-Maitland and Lynch 2020; Fiesler and Proferes 2018; Golder et al. 2017; Klassen and Fiesler 2022).

While there is limited research on the ethics of utilising public document data, extensive literature exists on the ethics of publicly made statements. This article explores the ethical implications of indirectly involving children in research by analysing court judgments, drawing upon existing literature on social media research. It aims to identify key areas that necessitate special attention from researchers and offer insights into effectively navigating ethical challenges. Using the Norwegian judiciary system as a primary example, given that Norway has more CPS cases appearing before the ECtHR than the rest of Europe combined (Helland and Hollekim 2023), the article contends that the discussion is applicable to all public documents where children are indirectly involved.

Identifying critical areas of concern

Previous research has identified four critical areas of concern when researching social media data (Townsend and Wallace 2016): (1) Private or public data; (2) Consent; (3) Anonymity; and (4) Risk of harm to research subjects. These four areas, which will lay the foundation for the article’s discussion, will be presented in the following section. Moreover, each concern will be placed in the context of researching court proceedings.

Private v. public

Research ethics cannot be ignored just because the data is seemingly public (boyd and Crawford 2012). Social media is created for sharing information. When creating an account to participate in the online world, users are often obliged to provide private information they otherwise would not divulge, thus forsaking some of their privacy (Barrett-Maitland and Lynch 2020). Where the data is found may influence ethical considerations, however (Townsend and Wallace 2016). There is a difference between using hashtags on Instagram and Twitter, broadcasting opinions to everyone, and stating an opinion in a private, password-protected online forum. Whereas hashtags may indicate a willingness – and perhaps a desire – to reach a broad, public audience, the opposite can be said for password-protected forums. Thus, the password-protected forum may be considered a private sphere, even though it is publicly available for all who register on the platform (Townsend and Wallace 2016). There is a balancing act between private and public data online; that the data is available does not automatically entail that it is public (boyd and Crawford 2012). In certain situations, arguments in an online discussion may sway between public political statements and private sensitive (Stang 2018), indicating that the participant has different opinions of whether the discussion is public or private. It may also be a question of knowledge: some seemingly do not fully understand that all online discussions are public – even in an ostensibly private sphere.

Court judgments are publicly available. Those familiar with judgments know that the judgments will be posted online. However, in child protection cases, court proceedings are often the biological parents’ last hope and resort to solve the situation and keep the family united. Private information will be relayed throughout the proceedings as the court considers all sides of a case to give a correct ruling. Unlike social media, individuals cannot opt not to share information as it will hurt their case. Thus, the case will provide an open window into the private sphere of someone’s family and home. To add to the concern, in most situations, biological parents bring the case before a court, not their children. Thus, the children are indirectly involved in a public proceeding which concerns them without being a direct part.

Informed consent

As a rule of thumb, consent should be the general standard in research involving human beings (NESH, 2021, 15). However, there are exceptions: if obtaining consent is impractical or impossible or if there is no direct interaction between the researchers and those involved, consent may be waived (NESH, 2021, 18). If that is the case, researchers have a responsibility to exercise caution and uphold the freedoms, rights, and human dignity of the participants. They must explicitly evaluate the advantages and disadvantages of the research and acknowledge the absence of alternative approaches (NESH, 2021, 18).

In social media research involving large quantities of data, such as data mining on Twitter, researchers (mostly) do not gain consent from tweeters (Fiesler and Proferes 2018). However, research has shown that when asked, the tweeters emphasise that they do not wish for their tweets to be used in research without their knowledge or consent (Fiesler and Proferes 2018; Klassen and Fiesler 2022).

The research on people’s attitudes towards being unaware participants in research emphasises the importance of adequately addressing the question of consent. Researchers have repeatedly discussed when consent is obligatory and when it can be waived, depending on the origin of the online data (see, e.g. Elgesem 2015; Klassen and Fiesler 2022; Mikal, Hurst, and Conway 2016; Stang 2018). In short, there seems to be a consensus that whether the data is considered private or public is critical. Still, there are disagreements regarding where each researcher believes the border between public and private online data to be drawn.

There is seemingly no clear answer regarding when informed consent should be applied to social media research. The angle of approach, the research question, and the origin of the data may all play a role in assessing the necessity of consent. Similar assessments can be used for discussing when informed consent would be obligatory when analysing child protection judgments. Moreover, people’s attitudes towards being part of a research project without their knowledge or consent in social media research will, most likely, be directly transferable to other research areas, such as analysing court judgments. For research on CPS, it is not the children who give consent; it is either their biological parents or guardians. This means that even if adults consent, the child the case revolves around does not (cf. Mathews 2023).

Anonymity

There is a discussion on whether social media data can be completely anonymous. Even when using aggregated data, there is a risk of reidentification, which entails that rewriting statements is insufficient to secure anonymity (Sormanen, Rohila, and Lauk 2016; Townsend and Wallace 2016; Williams, Burnap, and Sloan 2017). As previously mentioned, it would probably be more accurate to state that the data is deidentified rather than anonymous. Nevertheless, research has shown that, given a choice, participants prefer to be anonymised when participating in social media research (Fiesler and Proferes 2018; Klassen and Fiesler 2022).

Similar conditions apply to court judgments. Even if the judgments are anonymised, there is a theoretical possibility to reidentify the parts in the case. For judgments from the ECtHR, the case is often named after the applicants’ surname, e.g. Johansen v. Norway (1996) and Strand Lobben and others v. Norway (2019). Other cases are deidentified using the applicant’s initials, such as A.S. v. Norway (2019) and K.O. and V.M. v. Norway (2019). In both situations, identifying the applicants are easy for those with computer knowledge. The children are anonymised and thus not mentioned by name in the judgments. However, the children can risk being identified through the connection with their biological parents. This effectively means the children are deidentified with a potential risk of being reidentified in a case in which they are often not a direct part.

Risk of harm

Research on sensitive areas and vulnerable groups or individuals requires extra consideration for the potential risk of harm to those involved in the research (Townsend and Wallace 2016). The risks vary in severity, from undermining people’s trust in research to more severe risks, such as potentially stigmatising a whole group or poor anonymisation (Stang 2018). Elgesem (2015) argues that the risk of harm may steer whether consent is necessary. Consent is required if there is a risk of harm, and if there are no risks of harm, no consent may be considered. As mentioned above, the risks do not exist only on an individual level; there is a risk of stigmatising entire groups of people when researching social media data (Fiesler and Proferes 2018).

When considering the risks of harm, it is important to remember that social media data may present different angles to research. Kvalnes (2020) differentiate between ‘do-good ethics’ and ‘avoid-harm ethics’. He argues that for do-good ethics, social media should contribute to empowerment, transparency, employee engagement, sharing of knowledge, driving positive change and prosocial behaviour. For avoid-harm ethics, he emphasises that the use of social media should not contribute to loss of integrity, harassment, discrimination, trolling, fake news and destructive politics. Kvalnes highlights that research calls for a balanced view of the two different views on ethics.

There seems to be a consensus amongst different ethical guidelines and researchers that it is balancing different views, which is the critical factor in determining the risk of harm. Mathews (2023) discusses general principles of national research ethics in the USA, Canada and Australia. He argues that justifying any risk of harm requires the benefit of research. Moreover, the research design must be designed to minimise the risk of harm, and the researcher is responsible for the participant’s welfare. Similarly, NESH art. 28 states that it is the researcher’s responsibility to ensure that participants are not subjected to harm. However, the guidelines also state, ‘Research can impose participants with less severe or reasonable burdens, but this must be weighed against the research's benefit to society and value for those who participate.’ (NESH, 2021, 28).

Children require extra protection (NESH, 2021, 17). There is always a particular need to balance the benefits and harm associated with children’s participation in research (Backe-Hansen 2016). The balance between benefit and harm entails that the greater the sensitivity of the research area and, thus, potential harm, the greater the benefits needs to be. Harm should never outweigh benefits. Similarly, NESH art. 26 dictates that ‘the consideration of burdens on individuals outside the research project should be balanced against the consideration of the research's critical function’.

It is not only the immediate risk of harm that needs to be considered. Children in child protection may have experienced traumatic incidents (cf. Aakvaag and Strøm 2019; Kletzka and Siegfried 2008). If research is published and the child becomes aware of the research, it may cause them to relieve past experiences, adding to their burden and, consequently, increasing the risk of harm.

Overall, the risk of harm should always be crucial when conducting research. The utilisation of public data, such as social media data and court judgments, does not absolve the researcher from these responsibilities. Even when analysing public data that does not require consent, the research can still potentially impact individuals.

Discussion

This article sets out to discuss the ethical implications of indirectly involving children in research based on public data. Four critical areas of concern have been identified, all somewhat interlinked. The discussion is divided into two parts. First, there is a general discussion on the implications of indirectly involving children in research based on the four previously mentioned concerns. Second, building on the general discussion, the article will discuss different approaches to creating an ethical research design.

General discussion

Even though court judgments are public, they address individuals at their most private; the more sensitive matter, the more private the judgment (Backe-Hansen 2016). To issue a care order and remove a child from their biological parents is perhaps the most intrusive measure domestic authorities can do to a family. The ECtHR has established that a care order is, per definition, a violation of the biological parents’ human rights unless certain requirements are met (App no 22430/93, 1998, 51; App no 4547/10, 2012, 134). In short, these requirements focuses on the biological parents’ ability to care for the child. If they cannot care for the child, the child will be placed in public care. However, discussing the capacities and abilities of individuals at their most vulnerable is a very sensitive matter and, thus, very private. Even more, in a court setting, personal and sensitive information about individuals is shared, which affects the privacy of those involved (cf. Barrett-Maitland and Lynch 2020). The privacy issue may point to child protection judgments being private, even though the data itself is public (cf. boyd and Crawford 2012).

Despite care order cases revolving around a child’s well-being, the child is seldom directly involved. Nevertheless, the court shares and assesses information about the child and the child’s needs, such as e.g. traumas, reactions after meeting the parents, and the child’s development (see Helland and Hollekim 2023). Personal information about the child is another reason why court judgments may be considered private even though the data is public.

There is a clear difference between social media and court judgments. In social media, the participants themselves post the information online. In court proceedings, a case is only exempt from the public in special circumstances. Thus, court proceedings may be considered more private than social media by default.

Even though there are clear indications that public court judgments should be considered private, there are grey areas. The angle of approach, or research questions, may be crucial in determining which actions to take. The same goes for the question of consent. If the data is deemed private, informed consent is obligatory (Elgesem 2015). However, as children cannot consent, their guardians must do so. In care order cases, the role of the child's guardian is discussed. In most cases, the biological parents retain parental rights even though the child is in public care. This raises the question of whether it is the biological parents or the publicly appointed guardians who should consent to the research. If the biological parents are to consent after being subjected to the harrowing experience of potentially losing their children, their frame of mind can be questioned. If the legal guardians are to consent, the biological parents may oppose the decision, which may cause additional stress for both the child and their biological parents. On the whole: the question of consent on behalf of the child is a slippery slope which the researcher must tread with care.

In care order proceedings, the involved parties are, in most cases, the biological parents and the domestic authorities. If the research question is framed in a manner where it is the biological parents who are the participants, the child would potentially still be involved through the data material – even though they are not a direct part. Suppose a researcher deems it necessary to have informed consent to conduct the research. In that case, there is a possibility that the child should give assent¹ in addition to consent from the guardians (Felzmann 2009; Miller, Drotar, and Kodish 2004). Again it is dependent on the angle of approach to the research. If a researcher concludes that consent from the parties in the court proceedings is enough to cover any ethical concerns, NESH art. 26 clearly states that the potential burden on those outside the research area should be considered.

The ethical implications of directly or indirectly involving children in research is a balancing act (Backe-Hansen 2016). NESH art. 17 states that children need extra protection. However, even if children need protection, there is the question of benefits and the greater good. From a utilitarian point of view, the potential benefits, or the greatest good, for the highest number may outweigh the negative consequences of the few (Mill 2000). Even though it seems like a highly sensitive area and an ethical minefield, research involving children, both directly and indirectly, may be necessary for the greater good. As mentioned in the introduction, Norway has been under scrutiny by the ECtHR. The Norwegian Supreme Court has acknowledged that the practices in Norwegian child protection services require change (HR-2020-661-S 2020). Moreover, Norwegian authorities have repeatedly been criticised for their attitudes towards biological parents when a child is taken into public care (NIM 2020). Thus, research on care order proceedings may benefit many families in years to come.

It is unavoidable that some research areas of high tension or sensitivity require extra attention to ethically conducting the research. NESH art. 23 is based on the echr art. 8, stating that researchers should respect individuals’ private and family life. Care order cases concern art. 8 of the echr and, consequently, NESH art. 23. This does not mean that it is near impossible to conduct ethical research on care orders. It means that there must be certain guiding ethical principles of which the research must follow.

The guiding principles may bear a resemblance to a deontologist’s perspective. Deontologists argue that a researcher must abide by a particular set of rules for the research to be considered ethical (Alexander and Moore 2021). If guiding principles are in play, and the researcher follows them meticulously, the research is ethical. However, for a researcher to follow an inflexible procedure when dealing with sensitive areas is, at best, challenging. In stressful situations, every person reacts differently. What may seem ethically right in one situation may not be transferable to another case. Every person and every case is different. In short, qualitative data, such as court documents, require a degree of flexibility from the researcher (Bratberg 2017). However, flexibility is not limited to qualitative data analyses, the same principle applies to quantitative coding and analyses. If a researcher codes the data to conduct quantitative studies of the material, the data is subjected to the researcher’s discretion (cf. Tamutienė 2018). All in all, flexibility is, to a varying degree, present in research on both social media and court judgments. However, flexibility does not automatically render the research unethical. It is the angle of approach which is the critical factor.

Regardless of which approaches a researcher takes, minimising the risk of harm to those affected by the research is a given. Research has shown that, when asked, individuals who post things online prefer to be anonymous in a research setting (Fiesler and Proferes 2018; Klassen and Fiesler 2022). It would not be judicious to state that these opinions, most likely, are transferable to individuals involved in court proceedings. Complete anonymisation of individuals engaged in discussion online, be it through social media or court judgments, is nearly impossible (Williams, Burnap, and Sloan 2017). However, there is always a possibility to deidentify the participants. If participants are deidentified, aggregated data must be used to minimise the risk of reidentification. Online search engines have little difficulty finding the quote’s origin if direct quotes are included in the research.

Engaging ethical review panels in research design can serve as a means to reduce the risk of harm. However, Sellers, Samuel, and Derrick (2020) discovered that these panels often struggle to keep pace with the rapid developments in the field of social media, primarily due to the absence of institutional and professional guidelines. Nevertheless, despite these challenges, ethical review panels may potentially be a valuable tool for ensuring that researchers thoroughly consider all aspects of their research design.

The risk of harm is the overarching concern when researching public data. The researcher is responsible for ensuring that those involved in the research, both directly and indirectly, run no harm from being part of the research (NESH, 2021, 28). It is important for the researcher to also consider the potential future harm for the children, should they be aware of the research. Public data does not exempt the researcher from these considerations. As before, the angle of approach determines whether the potential harm outweighs the potential benefits of research.

Ethical consideration when constructing a research design

As mentioned, the angle of approach is crucial in determining which steps to take to ensure the research is conducted ethically when analysing social media or court judgments. The researcher must ensure that the research design addresses all relevant concerns. This article has presented four issues which may act as a guide towards creating an ethical research design.

Court judgments are, as discussed, both private and public. The data itself is public, but the angle of approach may cause the researcher to consider the data to be private. Stommel and de Rijk (2021) argue that in discussions on ethics in social media data, general claims of ethics, such as ‘the data is publicly available', should be avoided. Each research project should concretely discuss the ethics of the situations at hand. Parallels can be drawn to discussions on ethics and court judgments. Each judgment should be assessed on its own merits, and general presumptions should be avoided.

Regarding consent and assent: if children are not directly involved in the research, assent would likely be redundant, even if consent is needed from other participants (such as the biological parents) (Mathews 2023; Miller, Drotar, and Kodish 2004). However, if the research topic involves children, e.g. researching the relationship between biological parents and children in care order proceedings, the child should be included in the consent to the same degree as the biological parents. If the research focuses on the biological parents, e.g. the biological parents’ reaction after a care order, consent or assent from the child is more debatable. What is not debatable is the researcher’s awareness of unintended effects and negative consequences for the child who is indirectly involved in the research (NESH, 2021, 26). The researcher must balance the potential negative consequences for those involved against the potential benefits of the research (Mathews 2023; Stang 2018).

As previously discussed, complete anonymity is difficult to achieve for online data. For court judgments, anonymity varies. In Norway, judgments are available at Lovdata.no. Before publishing the judgments, all sensitive information is removed from the judgment. However, although Lovdata anonymises the judgments, most proceedings are open for public. Thus it is possible to identify those involved. ECtHR judgment involving Norway is also published and deidentified at Lovdata. However, there is a link in the judgment to the original judgment in the ECtHR’s database, which is not deidentified, thus rendering the deidentifying process in Lovdata pointless. It does give an extra obstacle for those who wish to reidentify the judgments, but nothing that is not easily overcome. Even though the parties in a case may be identified, the children’s names are kept out of the judgments. Nevertheless, the child is identifiable to those who know the families in question.

Stang (2018) discusses the link between anonymisation and public-private data. She argues that, when researching a protest group against child protection services on Facebook, anonymisation of the participation is sufficient to protect the participants’ privacy. She also emphasises that by not directly engaging with the participant and receiving consent, their considerations on whether the data is public or private, and the feelings of being ascribed to a given group, remain unknown to the researcher. Stang’s arguments are comparable to those favouring deidentifying court judgments if consent is not given. Taking necessary steps to protect the identity of the participants minimises the ethical concerns regarding consent. However, there is a balancing act regarding the likelihood of reidentification and, thus, possible harm.

For judgments from the ECtHR, names are known. Several of the judgments against Norway have been in the media worldwide (Lohne, Ege, and Norman 2015; Mæland 2015). The potential risk of harm does not dissipate just because a case is known. One of the most prominent child protection cases from the ECtHR in the last 20 + years is Strand Lobben and others v. Norway (App no 37283/13, 2019). This case is being researched and analysed domestically in Norway and internationally (Acconciamessa 2020; Köhler-Olsen 2019; Søvig and Tjelmeland 2019). It is also a case which has received substantial media coverage. At the time of the ECtHR judgment, the child was adopted by his foster parents. Even though the research revolves around the judgment, the Norwegian child protection services, and the ECtHR, the ethical concerns regarding the risk of harm to the child – who is indirectly involved in the research – should be considered. However, there is a question of the greater good. Strand Lobben and others has led to changes in the Norwegian child protection system (Barne – og familidepartementet 2023). The research on child protection services and the process leading up to the care order may help other children.

For the child in question in Strand Lobben and others, their adoptive parents have not consented to the research or frequented media. The biological mother, on the other hand, who is no longer the child’s legal parent, has. Strand Lobben and others is a prime example of a child being indirectly involved in social media, media, and other public documents. When discussing the risk of harm in care order cases, the burden the potential harm imposes on the child should always be discussed against the potential benefits of the research. For care order cases, the child has been through a harrowing experience. It is crucial to keep in mind that there is a reason why the child has been removed from the biological family. In many cases, children in public care have experienced assault, drug use, or other misconduct by their parents (Tregeagle, Moggach, and Trivedi 2019). Researching these cases may cause the child to relive past traumas, which the researcher should consider when assessing the risk of harm.

Conclusion and final remarks

The angle of approach is essential when researching sensitive areas. The main challenge in child protection cases is that the child is not necessarily a part of the research but rather an attachment to the biological parents. Consent and assent may help remediate the risk of harm in researh. Thus, the risk of harm should steer whether consent is needed for public data (Elgesem 2015). However, even if consent is given, the researcher should carefully consider the angle of approach regarding the inclusion of the child in the research.

For cases of sensitive matters, such as care order cases, the researcher needs to remember that there is a past for those involved. If children are not directly involved in the research, it may still burden a child if the child becomes aware of the research.

NESH art. 26 clearly states that the researcher should be aware of unintended effects and negative consequences for those indirectly involved in the research, with a special emphasis on private and close relations. Thus, the researcher’s focus must not only be on those directly involved in the research but also on those who may be affected by the research, with extra care for children (NESH, 2021, 17).

Ethical considerations will also be a balancing act between the risk of harm and the greater good. For research involving children, the weightiness of the potential risk of harm to the child should be an everpresent consideration. This also applies to research where the child is only indirectly involved.

When analysing public data, such as court judgments, the balancing act should start with the research design. By considering the four concerns raised in this article for both those directly and indirectly involved in the research and designing the research around them, the potential ethical challenges tied to the research are minimised.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Notes on contributors

Trond Helland

Trond Helland is a researcher at the Institute for Health Promotion and Development at the University of Bergen. He adopts a social science approach to studying case law from domestic and international courts, focusing on the development of children’s human rights, decision-making, child protection, and the child as a subject of rights. His methods are innovative, integrating theories from various disciplines to provide a broader understanding of his research topics. In September 2024, he has been invited to be Norway’s keynote speaker at the Nordic Congress on Child Welfare.

Notes

1 Legal consent has an age-limitation. If the age requirement is not met, the child assents rather than consents. The child’s assent comes in addition to the guardians consent.