A court file analysis of child protection cases: What do children say?

Correspondence Nicole Kratky, Department of Psychology, Technische Universität Darmstadt, Alexanderstraße 10, 64283 Darmstadt, Germany. Email: kratky@psychologie.tu-darmstadt.de Abstract Children's participation in legal proceedings affecting them personally has been gaining importance. So far, a primary research concern has been how children experience their participation in court proceedings. However, little is known about the child's voice itself: Are children able to clearly express their wishes, and if so, what do they say in child protection cases? In this study, we extracted information about children's statements from court file data of 220 child protection cases in Germany. We found 182 children were asked about their wishes. The majority of the statements found came either from reports of the guardians ad litem or from judicial records of the child hearings. Using content analysis, three main aspects of the statements were extracted: wishes concerning main place of residence, wishes about whom to have or not contact with, and children granting decision-making authority to someone else. Children's main focus was on their parents, but others (e.g., relatives and foster care providers) were also mentioned. Intercoder agreement was substantial. Making sure that child hearings are as informative as possible is in the child's best interest. Therefore, the categories developed herein might help professionals to ask questions more precisely relevant to the child.

Recently, some interview-based studies have examined the participation of children in family court proceedings and investigated how children feel while participating. They often did not feel comfortable in front of the court, nor did they feel listened to despite being heard (Block, Oran, Oran, Baumrind, & Goodman, 2010;Cossar, Brandon, & Jordan, 2016;McLeod, 2007). Nevertheless, children appreciated being active participants in proceedings (Mitchell, Kuczynski, Tubbs, & Ross, 2010;Thomas & O'Kane, 1999).
However, little attention has been paid to the child's voice itself.
Some studies have focused on the children's evaluation of being a participant and asked them questions on this topic after the court decisions were taken (Block et al., 2010;Dillon, Greenop, & Hills, 2015;Holland & O'Neill, 2016). Others have focused on the children's wishes by investigating reports by guardians ad litem (Vis & Fossum, 2013) or used case vignettes evaluated by child protection workers (Berrick, Dickens, Pösö, & Skivenes, 2015). Murray and Hallett (2000) investigated children's statements from judicial child hearings regarding what should happen to them. They found children speaking more when they were asked open-ended questions, but they did not systematically analyse the content of their statements. Consequently, it remains unclear what precisely children say while they are participating in child protection cases. Given that studies using court file data are very rare, the major objective of the present study was to investigate the content of children's statements by using data from court case files. By exploring these, this study aims to advance the knowledge of how children use their voices in court. We explored whether children express their wishes in a clear way, and if they do, to what or whom wishes are related.

| Participation of children in child protection proceedings
Child protection proceedings differ from other court proceedings insofar as a third party-the child-is involved (Dettenborn, 2010). In Germany, the Act on Proceedings in Family Matters and in Matters of Non-Contentious Jurisdiction (FamFG, December 2008), governs the child's participation. The Act states that an in-person hearing with the child is necessary if the child is at least 14 years old (Section 159, 1). Younger children shall be heard "when the preferences, relationships, or the desires of the child are significant to the decision or when an in-person hearing is otherwise indicated" (Section 159, 2). German family judges have stated that not hearing the child's voice could compromise their work (Rohmann, 2012). In German law, the court must appoint a guardian ad litem, who "determines the interests of the child and shall assert these in the court proceedings" (FamFG, Section 158, 3 and 4). The guardian ad litem must be appointed as early as possible to strengthen the position of the child among the adult participants in the proceedings. This guardian needs to have a legal, pedagogical, or psychosocial education, with an additional qualification in family proceedings, developmental psychology, and communication skills (Association of Guardians ad litem, 2012).
The current focus on the autonomy and participation of the child constitutes a rather new development. Looking back in history, the child's voice had not been regarded as worth of being respected for a long time (Parr, 2005). In 1980, listening to the child in legal proceedings was officially implemented in German law, and the judge is now required to involve the child if his or her preferences, attachments, and voice are of great importance for the court decision ( §50b Act on Voluntary Jurisdiction [FGG], now §159 FamFG, updated in 2018).
For a detailed description of the German child protection system and the procedures of child hearings, see Haug and Höynck (2017) as well as Salzgeber and Warning-Peltz (2019).

| The child's voice: Definition and assessment
The child's voice is the expression of the child's will. From a psychological point of view, the will is understood as the "age-appropriate, stable and autonomous orientation of the child towards pursued, personally significant target states" (Dettenborn, 2010, p. 66). The more attractive a target state is for the child, and the more likely it is to become reality, the higher the likelihood that he or she will prefer this target state. In most child protection cases, the child is asked directly about what he or she thinks and wants (Dettenborn, 2010). By default, the guardian ad litem talks to the child in person to explore the child's wishes and expectations regarding the relevant judicial questions. The guardian ad litem decides the number of conversations and then produces a report about the child's voice und welfare for the court (Association of Guardians ad litem, 2012; Dettenborn & Walter, 2015). The process in Germany is comparable with the practices in England and Scotland, as family judges then decide about the necessity of hearing the child's voice again by themselves (Raitt, 2004;§159 FamFG). If the child is asked by the judge in person, following §159 FamFG, the hearing takes place in the judge's room and not in front of all parties. If judges need further expertise, they may decide to commission a psychological expert who will also assess the child's voice (Rohmann, 2012).
However, the approach of directly questioning the child is somewhat controversial. Critics have debated whether a child has a direct understanding of his or her wishes and whether he or she is able to verbalize them clearly (Behrend, 2009;Shier, 2001;Spyrou, 2011).
Following Dettenborn (2010), hearing a child directly is unproblematic if it is done appropriately. First, this means establishing a proper setting for the child to talk, including a pleasant room and a trusting relationship (Archard & Skivenes, 2009;Benedan, Powell, Zajac, Lum, & Snow, 2018). Second, the child should be interviewed using open questions and without the pressure of specific expectations in order to avoid issues of suggestibility (Behrend, 2009;Benedan et al., 2018;Murray & Hallett, 2000). Third, the child needs to be informed about what will happen in the court proceedings, using language that is suitable for him or her (Archard & Skivenes, 2009;Raitt, 2004). The content of the child's oral testimony is then brought into the child protection case in an indirect way. The adults who interviewed the child inform the court of what the child said. What therefore needs to be considered is that the adults present the content of the child's statements, which can be affected by their own perception of the child and their personal interpretation of the child's words (Hunter, 2007;Spyrou, 2011;Zumbach, 2017).

| The present study
Being heard is of significant importance to children. Recently, the child's voice has increasingly been a focus of research. However, far too little attention has been paid to this topic as of yet. Sample sizes in existing studies are typically small, and original court files have rarely been investigated. To close this research gap, the present study analysed children's voices in a large sample of 220 family court proceedings. To investigate in detail what the children said, the content of the children's statements in the course of their child protection cases was extracted from the court files.
In doing so, we were able to analyse the actual views of the children while the cases were still open. We thus avoided the major issue of memory error, which comes along with a retrospective study design often used in this field of research (Spyrou, 2011;Widom, Raphael, & DuMont, 2004). We aimed to establish a coding scheme to conceptualize the children's voices. Therefore, we took advantage of a qualitative approach, which is useful in a scarcely investigated research field. We used content analysis (Elo & Kyngäs, 2008) and then analysed the frequencies and intercoder reliabilities of the categories.
To sum up, the study aims to investigate (a) what do children say when they are asked questions during child protection proceedings and (b) how often the observed categories of statements can be found in the studied cases. A combination of qualitative and quantitative research designs was used.

| Sample
To analyse the child's voice in child protection proceedings, court file data were examined. The court files were acquired after permission was granted by the Ministries of Justice of the federal states of Germany. From the 238 court files provided, 220 cases met the criterion of suspected child maltreatment, following §1666 of the German Civil Code, declared in January 2002. This paragraph includes the obligation of the state to intervene if parents are not able or not willing to fulfil their duty of parental care in the best interests of the child and if the well-being of the child is in danger. Judges must rule on the severity of the child maltreatment and then, if necessary, arrange homebased support for the family or decide if removal from parental custody is necessary. Figure 1 gives an overview of the children in the proceedings investigated. The total number of children in the families concerned was 503. Of these children, 343 children were directly affected by the proceedings, and 182 were asked about their wishes. For these children, it was possible to extract their voices from the court files, and results will be reported herein. These children showed a mean age of 9.3 years (SD = 4.5 years), and 54.4% of them were girls.
For a more detailed description of the sample, see Kratky and Schröder-Abé (2018).

| Development of the coding system
The results reported herein were part of a larger research project. So far, one paper has been published (Kratky & Schröder-Abé, 2018), using a different set of variables and data analyses. A coding system was developed, which was composed of four main categories: family characteristics, parental mental health status, child maltreatment information, and case characteristics (for a more detailed description, see Kratky & Schröder-Abé, 2018).
The child's voice was a subcategory of case characteristics. All quotations in the court files dealing with the child's voice were extracted word for word, including the information about from which type of document in the court file it was extracted. Some case files included verbatim accounts of what the children said.
Other cases did not, but the child's voice could be extracted from interviewers' reports. To gain an understanding of the voices' content, the quotations were investigated using content analysis. For a good overview of this method and its approaches, see Elo and Kyngäs (2008).
All quotation data were coded three times by three different coders. Classifying one statement into several categories was allowed for all categories, except for the category coded in Table 1, which classifies whether a statement was recorded in the documents and how it was classified in general. In the first round, the F I G U R E 1 Description of the sample categories were built using the inductive approach of content analysis. This approach is suggested for research fields in which little is known about the phenomenon under investigation. The categories were directly formed from the text material. Quotations sharing the same meaning were summarized into the same category (Krippendorff, 2013). In the second round, the material was coded using a primarily deductive approach. All categories from the first round were known to the second coder in advance. Creating new categories was permitted if the existing categories were not sufficient to represent the text material in the coders' view. This is required in content analysis to achieve a good fit of the categories to the text material (Mayring, 2000). The third round of coding was conducted to calculate intercoder agreement, and because new categories were established in the second round: the main category "authority to decide," the subcategory "wish for contact with the foster care family," and, concerning wishes for where to live, the subcategories "family of friends," "not father," and "not both parents." For assessing intercoder agreement, Cohen's kappa coefficients were calculated using the ratings from the second and third round (see Tables 1 and 3-5). Results showed a substantial to (almost) perfect agreement for most of the categories. For three subcategories, a moderate agreement was found (Hallgren, 2012). The results presented in this paper are based on the third round of coding. All quotations displayed were translated from German to English by the authors, where (…) are used to mark omissions and [...] are used to mark additions by the authors. Quotations presented in the paper are derived from various children from the sample. Table 1 shows if and how the child's voice was recorded in the court files. Cohen's kappa was .818 for this main category of "recording."

| RESULTS
For 163 out of the 343 children, no information about their wishes could be found in the files. For seven children, the attempt to examine their voice was not successful. In these cases, documents contained information directly stating that these children were not able to phrase their wishes. For example, a guardian ad litem said of a 2-yearold girl: "Due to the age of the girl, it was not possible to talk to her appropriately about her parents and her home. Nevertheless, she said 'my mum is nice and my dad, too.'" Further, 13 children did not say anything about their wishes, despite being asked about. For what reason remained unclear. These were summed up in the "no statement made, but asked" category. The following quotation from a T A B L E 4 Results concerning the main category "wish concerning personal contact" (n = 182) All three categories will be subsequently illustrated.

| Wishes concerning main place of residence
Many of the children expressed wishes about where they wanted to live in the future (see Table 3). Nearly three quarters of the statements showed a specific wish as to where to live or where not to live.
Most of the children expressed the wish to live with their mother. For example, a 6-year-old girl announced that "she would like to live with her mother in the mother's home." A more detailed statement was found in the record of a court meeting and child hearing involving a 16-year-old girl: "After the [parental] separation, which was very stressful for her, the girl tested out where her place of residence could be.(…) She then spent a short period in a residential facility. There, she had the positive experience that her mother still kept in contact with her and was interested in her. Therefore, she decided to go back to the mother.(…) 'I want to keep living with my mother, because she is taking care of me and is no longer losing control of her temper.'" Other children expressed the wish to live with their father. For example, a 13-year-old boy, when questioned by the family judge, said, "I know that this is about whether we will go back to our parents; and I want to go to my father. In the meantime, I was there for a whole weekend and it was nice.(…) I would prefer to be with my dad (…), I do not want to stay in foster care, I would like to go to my father." In some cases, the wish to live in a foster care setting was expressed. A guardian ad litem said of a 3-year-old boy and his siblings that "the children feel comfortable in the foster home and said that they wanted to stay there.( … ) No child has enquired about the mother (…). In between, the boy was at the hospital and expressed the wish to go home. It turned out that by 'home' he meant the foster home. When I said goodbye to the children, they said that (…) they wanted to stay where they are." Furthermore, some children expressed the wish to live with either both parents, relatives or siblings (see Table 3). Concerning relatives, children mentioned their grandparents most often (seven out of 16); this was followed by a wish to live with an aunt or with their older siblings.
Besides statements about with whom they wanted to live, children expressed wishes about with whom they did not want to live (see Table 3). In most of the cases, the mother was mentioned, for A few children said that they did not want to live with their father, both parents, or relatives (all grandmothers; see Table 3). Some children gave reasons for their wish as to where to live or where not to live. This included, for example, the quality of the relationship, how everyday life is arranged at the residence, and previous experiences they had with their parents. With regard to the latter, we found a 6-year-old boy saying he wanted to live with his mother, "because the father had hit him and his mother," an 8-year-old girl mentioning that she had to do things alone because her mother "had no time for her," and a 3-year-old boy saying he wanted to live with his father, "because the mother doesn't want him anymore."

| Wishes concerning personal contact
Another main category included statements about with whom the children wanted to have contact (see Table 4).
The father was the person mentioned most often. A 2-year-old girl "(...) said happily, that she would like to see her father again soon," as written in the record of the child hearing. Other children expressed the wish to stay in touch with their mother, like the 12-year-old boy who said "it would be fine for him to have contact with the mother twice a week. However, in no case would he want to stay constantly in contact with her," as reported by a family judge.
Again, children also expressed with whom they did not want to have contact (see Table 4). One guardian ad litem stated, "[child's name, 8 years old] has been living with his aunt and his grandparents for eight years and has been feeling comfortable there. His aunt is his primary attachment figure. The topic 'mum' was unpleasant to him.
He did not want to talk about her. He also does not want to see her." In another case, a guardian ad litem mentioned that "he [5 years old] did not want to have contact with the father. He did not want to give reasons for that." In some cases, children justified why they did not want to see a particular person. For example, an 8-year-old girl said she "did not want to have contact with the father because of his alcohol problems." In addition to such illnesses, children also named the negative quality of the relationship and negative experiences in the past as reasons.
Several children (n = 24) mentioned wishes regarding contact and main place of residence. The most common combination (n = 15) was naming one parent to live with and stating whether or not they wanted to have contact with the other parent, such as a 13-year-old girl wishing to live with her mother and "(...) also wants to see her father regularly." The second most frequent combination was expressing a desire not to live with parents but wishing to have contact with them (n = 4), like the 13-year-old girl who said in the courtroom that she "can imagine living in sheltered housing, provided that regular contact with the mother could persist."

| Statements concerning authority to decide
A small number of children mentioned a person to whom they would like to give the authority to make relevant decisions on their behalf (see Table 5).
In four out of nine cases, the authority to decide was given to relatives. One guardian ad litem stated that "the 17-year-old boy mentioned being disappointed about the father's withdrawal.(…) He wishes custody to be transferred to his aunt ('I am happy if my aunt gets custody')." Another child, a 15-year-old girl, wished "to live with her older sister and also wished that she [the sister] should decide any important issues. Additionally, child maintenance payments and child benefits should be transferred to her sister." In four cases, children said that their mother should make the important decisions. "When asked who should be allowed to make the important decisions for him [4 years old]-his mother, or his father or both parents together-he answered that he would prefer his mother to do so." One 14-year-old boy wanted to assign his mother the decision-making authority regarding how to arrange contact with his father: "I think it's best if my mother decides whether we should visit dad. She knows better, and she can recognize if dad is feeling well or not." One 14-year-old girl expressed the wish that the youth welfare office deals with the authorities and manage any necessary paperwork, because "if the mother does this, it does not work and always takes a long time." This statement also contains the reason why the child does not want the mother to manage the paperwork.
This was the only reason given in this category of statement.
Three children expressed wishes concerning the decision-making authority in addition to wishes regarding main place of residence. In all of these cases, the same person was addressed by both wishes. As an example, see the statement of the 15-year-old-girl showed above.

| DISCUSSION
The present study was designed to investigate children's voices in a sample of child protection cases. Children's statements in the child protection proceedings were extracted from the case files and analysed. We found statements concerning three main areas: children told interviewers where they wanted to live, with whom they wanted to have contact, and who should make decisions for them. Some children also noted with whom they did not want to live or have contact.
The majority of the children's statements were derived from reports of guardians ad litem or family judges. Additionally, a smaller number of quotations was extracted from reports of psychological or pedagogical experts.

It is not surprising that the clear majority of statements concerned
where the children should live (Vis & Fossum, 2013). Child protection proceedings often centre around this question, and children were directly asked about this topic. The mother was the person mentioned most often, followed by the father, and then the wish to live in foster care. The fact that the statements concerning the mother by far outnumber those concerning the father can be explained by the high prevalence of single parenting in the present sample. On the one hand, the wish to live outside the family in a foster care setting might be surprising because it means a drastic intervention, changing the child's life, and interrupting the relationships the child is used to. On the other hand, this wish might be explained by the fact that we investigated child protection cases in which child maltreatment was suspected or present. The wish to live in foster care may illustrate the child's hope to be protected from further maltreatment. Recent research from England has demonstrated that children are able to see possible positive consequences of foster care (Selwyn & Briheim-Crookall, 2017). When investigating where the children did not want to live, mothers and foster care were mentioned most often.
Children also told the interviewers with whom they wanted to have contact, naming their mother and father most often. As the main caregivers and attachment figures, parents are the people most present in the minds of the children. The wish to have contact with the father was recorded with a slightly higher frequency than the wish to have contact with the mother. This dovetails with the wishes regarding the main place of residence, where the mother was named most often. If a child stays or wishes to stay with a certain person, it can be expected that this person will not be named in the contact category, because the contact is already implied. In these cases, we often found children saying that they would like to have contact with the other parent, the one they would not be living with. Continuity of relationships is known as an important aspect of children's psychological and emotional stability (Behrend, 2009), and so it is in child protection cases, too (Montserrat, 2014). In addition, having contact with both parents is, by German law, a right of the child and needs to be considered by the judge ( §1684 German Civil Code). We found that some children wished no contact or to have contact in a supervised way, presumably to be protected and to rebuild trust. Some children justified this with previous experiences they had with their parents. Apart from parents, other relatives and friends were mentioned by the children. This shows the complexity of personal relationships and highlights that children are aware of relevant relationships besides those to their parents.
In some cases, children determined who should make decisions for them. Relatives and the mother were named most often. Deciding where to live, and thus deciding in favour of one parent, might bring children into some conflicts of loyalty (Dettenborn, 2010;Hunter, 2007). It may also entail fear of saying something that might have negative consequences for one parent (Montserrat, 2014). Saying nothing or granting someone the decision-making authority could be taken as hints for the conflicts that children might face during child protection proceedings. When interpreting this category, one should bear in mind that the number of examples we found was very small.

| Strengths, limitations, and directions for future research
This study is one of the first to systematically analyse the content of the child's voice using a court file database. We had access to the documents the judges used for decisions, so we were able to avoid the subjectivity that comes along with survey or interview methods (Widom et al., 2004). Compared with samples investigated so far (Cashmore, 2011;Cossar et al., 2016;De Bortoli, Coles, & Dolan, 2013;Mateos, Vaquero, Balsells, & Ponce, 2017), our sample with 220 cases is quite substantial in this field. Our study combined qualitative and quantitative methods, which we deem appropriate due to the lack of research in this field so far. Coding was performed three times to adapt the categories as close to the data as possible.
With few exceptions, we found substantial to (almost) perfect intercoder agreement (Hallgren, 2012), substantiating the quality of our category system.
Besides the strengths of using court files, it also entails possible limitations. The court files included letters from youth welfare offices and attorneys, records of court sessions and child hearings, medical and/or psychological reports, and court orders. However, there might be more information that was not precisely recorded.
Thus, we cannot preclude an underreporting of information (Afifi et al., 2015;de Bortoli et al., 2013 Similarly to the family judges in the proceedings, the coders in our study had to interpret the children's voices on the basis of the information that was documented in the court files. Calculating intercoder agreement thus allows one to approximate the amount of objectivity when interpreting such statements (Spyrou, 2011;Zumbach, 2017).
The results of our study may thus suggest that different adults come to the same conclusion when analysing the statements. However, intercoder agreement is influenced by many factors, such as the material to be coded, the quality of the coding system, and the training of the coders. Thus, considerably more research will be needed to determine the accuracy with which the child's voice is brought into court proceedings.
It would be interesting to investigate why there were children who did not say anything despite being asked. Saying nothing stands in contrast to previous findings, that being heard was of higher importance to children than whether the final decision was congruent with their wish (Bessell, 2011;Cashmore, 2011;Thomas & O'Kane, 1999).
Reasons for saying nothing could include the feeling of not being taken seriously, loyalty issues, or not being able to express the own voice due to psychological development (Archard & Skivenes, 2009;Astington & Edward, 2010;Montserrat, 2014). Interviewers should be trained in how to interview children appropriately to deal with these possible issues. This would also avoid multiple interviews, which often results in stress for the children and might make them feel like they had done something wrong in the first interview (Block et al., 2010).
Being asked repeatedly can also produce contradictory statements, which we found in a small number of cases. On the other hand, repeated interviews or meetings are sometimes necessary for getting to know each other, building up a trusting relationship and assessing the child's voice (Archard & Skivenes, 2009;Benedan et al., 2018). This is especially relevant for guardians ad litem, who have more time with the child than the judges. Beyond this, it would be interesting to analyse the length of the children's statements and if the detail of the children's voices differs by the child's age, by the person conducting the interview or by the timing of the child hearing during the proceedings.
In practice, there are some requirements concerning the child's voice that the judge considers while judging. The child's voice needs to be stable, intensive, autonomous, and orientated towards a target (Dettenborn, 2010). Further research is needed to account for the varying manifestations of these voice characteristics. However, bearing in mind how other children expressed their opinions might help to ask relevant questions to other children. This might simplify the process for interviewers and especially for children. Creating an interview setting that is as informative as possible is in the best interest of the children, which itself is the overriding principle of child protection cases.

DECLARATIONS OF CONFLICTS OF INTEREST
None.