Contact with parents from childhood to adulthood – a longitudinal study of children in kinship care and non-kinship care

ABSTRACT Previous research on parental contact for children in foster care shows that contact frequency is significantly related to parent’s gender and placement type. Yet very few studies have explored the impact of gender and type of placement on parental contact over time. Based on longitudinal quantitative data from kinship care and non-kinship care placements in Norway, we analyse contact between children and their birth parents at three timepoints: childhood, adolescence, and adulthood. The study shows different contact patterns for children’s contact with mothers and fathers. For most of the children, the possibility of contact with the mother was established from the first timepoint and contact occurred rather frequently throughout childhood, adolescence, and adulthood. Regular contact between fathers and children in childhood, on the other hand, was less common, and this could develop in different directions: either towards children coming into contact with their father later in life or not at all. From a longitudinal perspective, we can say that the most significant changes evolved around children’s contact with their fathers, while contact with mothers was more stable.


Introduction
The aim of this study is to examine how placement type and gender are linked to parental contact for children who grew up in foster care (FC) in Norway and how this might change over timefrom childhood to adult life.From a historical perspective, traditional gender roles combined with higher mortality rates among men have involved that mothers often have had a more present role in children's lives than fathers (Sogner, 1990).With changing expectations towards fathers to be more involved at home, and extended life expectancy among men (and women) (Herlofson & Daatland, 2016), this has changed, and today most children in Norway grow up having frequent contact with both their mother and father.Statistics show that almost 80 percent of Norwegian children under the age of 18 live with both parentsa number which has remained stable over the past fifteen years (Statistics Norway, 2022).Moreover, over 90% of young adults (18-29 years old) have both parents alive, and contact with parents continues in adult life for most of them (Herlofson & Daatland, 2016).
Among the 25,000 children who yearly experience parents' separation, between 30% and 50% live more or less equally with both parents (Kitterød & Wiik, 2017).And while most children in separated families still have permanent residence with their mothers, most fathers continue to have regular contact with their children, also after separation.In Lyngstad and colleagues' study of fathers not living with their children (2015), only three percent had no regular monthly contact with their children.Little or infrequent contact was found to be more likely when fathers had poor health and low education and income.Moreover, no holiday and no physical monthly contact was more likely when children were older (between 15 and 17 years old) than when they were younger (younger than 10 years old).According to the authors, this might be because teenagers are more likely to decide for themselves how much time they want to spend with non-resident parents, and often prioritise other activities (Lyngstad et al., 2015).
This study focuses on children who grew up in FC, a group rarely represented in family studies and where children are separated from both parents.FC can generate contact patterns quite different from those observed in the general population, partly because contact is regulated by child welfare services (CWS) to avoid further neglect or abuse.

Parental contact for children in foster carethe Norwegian context
Whereas parents and children's rights to have contact have a relatively long history in Norway, dating back to the Children's Act of 1956 (Haugli, 2000), it was not until the 1980s that facilitating contact between children in care and their parents became an established practice (Haugli & Havik, 2010).Children's rights to have contact with their parents (and vice versa) is covered by the United Nations Convention on the Rights of the Child and the European Convention on the Human rights.The Norwegian legislation states that 'Unless otherwise provided, children and parents are entitled to access to each other' (Child welfare Act 1992, section 4-19 Right of access.Covert address). 1 In Norway, knowledge of contact frequency is primarily available from one research project where data was collected at two timepoints (Havik, 1996(Havik, , 2007)).Results from this project show that the percentage of children in FC having contact with at least one parent has increased over time, from about 50% in 1985from about 50% in to 81% in 1994from about 50% in and 89% in 2005from about 50% in (Havik, 1996from about 50% in , 2007)).Today, most children in FC in Norway have some form of contact with at least one birth parent (Stang & Baugerud, 2018).The growing percentage of children in foster care having parental contact might tell us something about a general development influenced by legislation.Yet, it tells us little about how the frequency of contact evolves on an individual level.The purpose of this study is to enhance our understanding of how the possibility and frequency of parental contact might change over time for children in FC.

The longitudinal changes in parental contact
How contact between children in FC and their parents changes through time is an issue which has received little attention in research.In Hunt et al. study, contact usually diminished or ceased five years after the care proceedings (2010).A similar finding can be found in Kertesz et al. (2022) who explored children's contact with family members within a 12-months period.Here, children on non-reunification orders were 4.5 times more likely to not have had contact with at least one parent in the 12 months preceding data collection.We cannot find any study exploring parental contact for children in FC from childhood to adult life.
A longitudinal study involves collecting data on the same individuals over several timepoints.Longitudinal research can explore change and stability over time and address research questions that are difficult to answer using cross-sectional research designs (Kalaian & Kasim, 2011).In FC research, studies with a longitudinal design have been emphasised as important to better understand parental contact with children (Sen & Broadhurst, 2011;Stang & Baugerud, 2018).Compared to children in the general population, children in FC have less evident contact patterns with their parents.Reasons for this include parents problematic and changing life situations (Johnson et al., 1995).The accessibility of parents deprived of the right to care for their children due to drug abuse, for example, can vary greatly through time.Hence, measuring contact at one point in time gives limited insight into the phenomenon.A longitudinal exploration can give insight into whether contact continues once children reach adulthood when they have more possibilities to choose for themselves and the relationship is no longer regulated by CWS.In this study, we use longitudinal data to test whether the frequency of contact between parents and their children in FC changes over time.We expected the frequency of contact to be rather stable while regulated by CWS, and to decrease when the children reached adulthood.

Contact and type of placement
While child welfare policy, guidelines, and legislation play a key role in facilitating contact, the type of placement is often thought of as important for contact frequency, for example, whether the child grows up in kinship care (KC) or non-kinship care (NKC).Most studies show that children growing up in KC have more contact with their birth parents, compared to children in NKC (Berrick et al., 1994;Farmer & Moyers, 2008;Holtan, 2002;McWey & Cui, 2017;Rowe, 1984;Taplin & Mattick, 2014).This finding can be explained by characteristics of KC placements: the foster parents and the child's parents often know each other well, making visits more accessible.Moreover, some kinship carers see themselves as being responsible for maintaining contact, whereas non-kinship foster parents often perceive this as a task for the CWS (Le-Prohn, 1994).This is partly confirmed in the study of Hunt et al. (2010, p. 78), showing high levels of commitment to preserve parental contact among kinship carers.
Not only does it seem to matter where children are placed, but also who the child is placed with.Being placed with maternal relatives seemingly increases the chances of remaining in contact with at least one parent, and a parent is more likely to have contact if a child is placed with their own side of the family (Farmer & Moyers, 2008;Holtan, 2002;Hunt et al., 2010).Although a few studies have found contrasting results when comparing KC and NKC (e.g.Strijker et al., 2003), we expected our study to find similar results as the majority of research: that children placed in KC would have more often contact with their parents, and especially with parents on the side of the family where they are placed.

Contact and parents' and children's gender
The parent's gender seems to be a significant factor related to contact frequency in all countries with available data on parental contact in FC.Previous studies have found that children are more likely to have contact with their mother than their father.Children are also more likely to see their mother more frequently (Farmer & Moyers, 2008;Havik, 2007;Hunt et al., 2010).For example, in a nationally representative study of children and youth (6-17 years old) in FC in the US, 52% of respondents had never contact with their father, while 22% had never contact with their mother.Furthermore, 59% had contact with their mother at least weekly, while only 29% of the sample visited their father with the same frequency (McWey & Cui, 2017).In Norway, Havik found that 76% of the children had contact with their mother and 47% of the children had contact with their father in 1994 (1996).In 2005, the percentage of children who had contact with their mother had increased to 83%, while for fathers it had increased to 56% (2007).Newer data collected by The Norwegian Foster Care Association 2 shows little change, indicating that findings from Havik's last study (2007) are still relevant.We therefore expected that most children in our study would have contact with their mother, while fewer children would have contact with their father.We also expected a higher frequency of contact with mothers than fathers.
Whereas parents gender seemingly is an important factor, the impact of children's gender on parental contact is less clear.In Hunt and colleagues' study (2010), fathers were more likely to lose contact with boys.Other studies, however, have not found the child's gender to be of significance for parental contact (Egelund et al., 2008;Havik, 2007;McWey & Cui, 2017).There has been far too little attention given to this question to draw conclusions about the relationship between children's gender and contact with the birth parents while in FC.One of the aims of our study was thus to examine this question in the Norwegian context.

Participants and procedure
The material for this article is drawn from a longitudinal study of KC, conducted in Norway between 1999 and 2015.Data collection occurred through three waves: in 1999-2000 (T1), 2006-2008 (T2), and 2014-2015 (T3).At T1, foster parents of children (aged 4-13 years old) living in court-ordered KC and NKC, were recruited for participation.In all KC families, foster parents were relatives of the child (see also Skoglund et al., 2022, p. 6).A total of 234 KC families in 104 municipalities were asked to participate, representing 98% of all registered KC placements in Norway in the year 2000.About 55% accepted to participate (n = 129).In the NKC group, 192 families from three counties representing three different regions in Norway were invited to participate.Ninety one of these NKC families (i.e. about 47%) agreed to participate.Foster parents completed questionnaires on several topics, including parental contact.We received answers at T1 for 141 children placed in KC and 113 children placed in NKC.
There was more than one foster child living in the foster home for 32 of the 220 foster families included in this study.In nine cases, the foster children were 'full' siblings; in 19 cases, they were 'half' siblings; and in four cases, the children were not siblings.The most common living situation before the placement was living with a single mother (46%).The second was living with both parents (26%).
At T2, we invited the same foster parents to answer a follow-up survey, except for two children who had changed foster home and where the new foster parents answered the survey.We received answers at T2 for 72 children placed in KC and 63 placed in NKC (response rate: 53.4%).At T3, children had become young adults aged between 19 and 29.Only 223 participants could be invited to participate, as some of the participants at T2 did not consent to be contacted again for the follow-up study.Recruitment at T3 had to go through the foster parents as we did not have the young adults contact information.Of the 223 possible participants, at least 52 did not receive our invitation (27 because their foster parents could not be traced, 20 because their foster parents did not want to disclose their name and address, four who had died since T2, and one who was in prison).Of the remaining 171 young adults, 76 participated in the study at T3, either by answering a questionnaire and/or by participating in interviews (response rate at T3: 44.4%).The Regional Ethical Committee and Norwegian Data Inspectorate approved the study.

Measures
In FC research, parental contact has been defined as intentional communication between children and their parents (Bullen et al., 2015).Contact can be direct or indirect (Sen & Broadhurst, 2011), including formal visits arranged by CWS, informal visits which may or may not have been supervised, telephone calls, and other types of communication.In a longitudinal study, parental contact becomes a complex issue as the types of contact available at one point can be very different from another.The child's age is an important reason for this.At the same time, technological developments over the past two decades have given new opportunities for communication.For these reasons, parental contact is defined as physical contact at the two first measurements, and more broadly at T3.We specify this below.
Variable names are identified using italics.At each timepoint, we used two variables to evaluate contact with the parents: Possibility for contact and Frequency of contact.Possibility for contact assessed the availability of the parent and was coded into four categories: parent is dead, parent is unknown or has never had contact, contact with parent is disrupted, and ongoing contact.We determined the Possibility for contact with each parent at T1, based on when the last contact with the mother/father happened.When the last contact had occurred more than two years ago, we registered the Possibility for contact as 'contact disrupted'.We updated the categories at T2 and T3 according to the changes reported by the foster parents or foster children.Frequency of contact measured how often the child was in contact with the parent and was coded into five categories: never, less than once a year, a few times a year, monthly, or weekly contact.This variable evaluates slightly different events throughout the study as the foster parents were asked at T1 and T2 how often the child had been together with his/ her parent (i.e.physical contact), while at T3, the young adults were asked how often they had been in contact with their parents (including physical contact, phone contact, emails, etc.) Type of placement was first evaluated as two categories: KC vs. NKC (Type of placement 2) and later recoded into three categories to further differentiate between children placed on the mother side vs. the father side (Type of placement 3: placed in KC on the mother side, placed in KC on the father side, placed in NKC).

Attrition analysis
We tested for a possible selective dropout from the study with a binary regression, using participation at T3 as the dependent variable.We examined whether gender, age, type of placement (Type of placement 2), Possibility for contact and Frequency of contact at T1 predicted participation at T3 (Table 1).Age and gender of the respondent were significant, with chances to participate decreasing for older participants (β = −0.104± 0.053, p = 0.049) and for boys compared to girls (β = −0.678± 0.287, p = 0.018).

Data analysis
We first examined how Possibility for contact changed through time using descriptive statistics and tested with chi-square tests whether the type of placement at T1 (Type of placement 2: KC or NKC) and the child's gender (Sex: male or female) were related to Possibility for contact.We used adjusted standardised residuals to conduct post-hoc tests and Bonferroni correction to calculate the adjusted P-values associated with the tests.
Second, we calculated descriptive statistics for the Frequency of contact for each timepoint, selecting only respondents having a living mother/father at that timepoint.This gives an idea of how often the respondents answering the survey and having a living mother/ father had contact with their parents.Lastly, we used multilevel modelling to test whether Frequency of contact changed through time (test for significantly different contact frequency between the three Timepoint: T1 vs. T2 vs. T3) and whether the type of placement and the child's gender predicted Frequency of contact.Multilevel modelling can be used to analyse longitudinal data, where the observations at each timepoint (level 1) are nested within individuals (level 2).Because some of the individuals were also nested within the same families (level 3), we also tried to fit models considering this third level, but those models were too complicated given our sample size and did not reach convergence.We also tried keeping only one sibling in our analyses when several siblings were living in the same foster home and run our models on a lower number of foster children (223 instead of 254) and then found the same results as the ones obtained with the full sample.Since most of those children were 'half' siblings, and often had different possibilities and frequencies of contact with their birth parents, we chose to present results obtained with the full sample.
We fitted linear mixed models with Frequency of contact as the dependent variable and an unstructured covariance matrix, which allows variances and covariances for each timepoints to be estimated freely.Models were fitted using the restricted maximum likelihood method.Mixed models integrate the available data at each timepoint, and do not exclude individuals that are missing data at a given timepoint from the analysis.They enabled us to test whether Frequency of contact changed through time and whether other independent variables of interest (e.g.Sex) were linked to Frequency of contact.We ran separate models for Frequency of contact with the mother and Frequency of contact with the father.The effect of the type of placement was first examined as two categories (Type of placement 2: KC or NKC) and afterwards as three categories (Type of placement 3) to test for differences between children that were placed on the mother side and children that were placed on the father side.

Results
Our total sample consisted of 254 children born between 1986 and 1995, where boys were slightly overrepresented (ca.55% of the participants, Table 2).About 55% were placed in KC, with placement on the mother side being more common than placement on the father side (ca.39% vs. 16%, Table 2).

Possibility for contact
A large majority of the children had contact with their mother throughout childhood (93% at T1 and T2) and young adulthood (72% at T3).None of the mothers were unknown, but a few children stopped having contact, most often because the mother died during the course of our study (Figure 1).Eight of the 253 children had a mother who was dead at T1 (ca.3%), while this was the case for 15 of the 76 youths that answered the survey at T3 (ca.20%).
The percentage of children maintaining contact with their father stayed rather stable, around 50-57% throughout the study.About 13% of the fathers were unknown/had never had contact with their child at T1 and T2, but this percentage decreased to about 4% at T3.On the other hand, the percentage of children who had lost their father increased throughout the study (from 8% at T1 to 21% at T3), but this increase was more gradual than the increase we observed for the mothers (Figure 1).
Chi-square tests of independence showed that there was no significant association between the Possibility for contact and Type of placement 2 (all P > 0.07).Possibility for contact was generally not related to the Sex of the child, except for contact with the father at T2 (χ 2 (3,135) = 10.6,P = 0.014).Examination of the adjusted standardised residuals showed that boys had disrupted contact with their father more often than girls (P = 0.011), while girls had more often ongoing contact with their father at T2 (P = 0.049).

Frequency of contact
First examining descriptive statistics for respondents having a living mother/father, we found that most of the respondents with a mother who was alive had contact with their mother at least once a month (49.9% at T1, 69.7% at T2 and 54.2% at T3).The percentage of respondents who never had contact with their living mother remained quite stable throughout the study: about 6% at T1 and T2, and nearly 10% at T3. Frequency of contact with a living father was generally lower than the one observed for the mother, but the percentage of respondents having contact with their father at least once a month evolved from 23.1% at T1 to about 36% at T2 and T3.The percentage of respondents having never contact with their living father remained quite stable around 38-39% at T1 and T2 but decreased slightly to 33% at T3 (Table 3).
Examining the results obtained from the mixed models, where all respondents were included (i.e. both those with living and non-living parents), we found that Frequency of contact did not change significantly according to Timepoint, neither for contact with the mother nor contact with the father (all P > 0.26, Tables 4 and 5).We found no effect of Type of placement 2 (KC or NKC) on the Frequency  EUROPEAN JOURNAL OF SOCIAL WORK of contact with the mother (P = 0.062, Table 4).For contact with the father, however, being placed in KC increased the Frequency of contact (P = 0.012, Table 4).When looking at the type of placement as three categories (mother side, father side, or NKC), we found that Frequency of contact with the mother was higher for children placed on the mother side (P = 0.001), and lower for children placed on the father side (P = 0.035), as compared to children placed in NKC (Table 5).Contact with the father, on the other side, was higher for children placed on the father side compared to children placed in NKC (P < 0.001, Table 5).However, there was no significant difference in contact with the father for children placed on the mother side vs. those placed in NKC (P = 0.35, Table 5).Girls also had more frequently contact with their father than boys (P = 0.010, Tables 4  and 5).

Discussion
The aim of this study was to examine how parental contact for children in FC is linked to placement type and gender, and how this might change over timefrom childhood to adult life.We now turn to discuss our findings in light of previous research on parental contact, with particular focus on what the longitudinal aspect of the study can tell us about change and stability in contact over time.
Placement type: Effect on frequency, but not possibility for contact In our study, there was no effect of KC vs. NKC on the possibility for contact.This suggests that there are not two different 'population of children' (e.g.those who have their parents available who are placed in KC vs. those who have no parents available, who are placed in NKC).The parents were as much available, for children in KC and NKC.
In line with previous research, we found an effect of placement in KC on the frequency of contact (McWey & Cui, 2017;Moyers et al., 2006;Rowe, 1984), with children placed in KC having contact with their father more often than children placed in NKC.When looking in more detail at this effect, it seems to be mainly related to KC on the father's side as children had more often contact with their fathers than children placed in NKC.Possible hypotheses to explain this finding include that the father's family may facilitate contact with the father, and/or that children might be placed on  the father side more often if the father is involved earlier in the child's life and thus more likely to maintain contact.Previous studies have found that type of placement matters also for children's contact with the mother (e.g.Hunt et al., 2010).Although there was overall no significant effect of KC vs. NKC on the frequency of contact with the mother in our study, we found an effect of the type of placement when separating between placement on the mother side, father side, or in NKC.Farmer and Moyers (2008), who found that children placed with relatives on the mother's side have more contact with their mothers, and children placed with relatives on the father's side have more contact with their fathers, describe this finding as 'unsurprising ' (p. 223).Implicit in this description is that children's parents have easier access to their children when the child is placed on their side of the family.What is perhaps less evident is our findings showing that children placed on the father side have less contact with their mother than children placed in NKC.This might be due to children being more likely to be placed on the father's side when the mother's life situation is particularly challenging.In such cases, the frequency of contact might be either reduced by the decision of the CWS, or because the mother doesn't show up to agreed contact times or perhaps because foster parents from the father side may impede contact to some degree (for example in case of conflicts between the two family sides).

Most children have contact with their mother, fathers are often absent
In a Spanish study (León et al., 2017), one-third of the children had no contact with either parent while growing up in FC.Some decades ago, we might have found similar results.However, given the increased importance ascribed to parental contact in Norway, along with the findings from Havik (1996,2007), we are not surprised to find that most children had contact with at least one parent while growing up.More specifically, the majority of the children in our study (independent of placement type) had contact with their mothers from childhood to adulthood (unless the mother was dead).In contrast, quite some fathers were unknown in childhood.The percentage of children maintaining contact with their father stayed rather stable, around 50%-57% throughout the study.This leaves about 43%-50% of the children who had no possibility of contact with their father from childhood to adult life, either because the father was dead, unknown, or because the contact with the father was disrupted.Hence, in a time when the 'absent father' was deemed outdated in Norway (Brandth & Kvande, 2003), fathers absence was a reality for almost half of the children in our study.While this percentage is substantial, the finding is not surprising, given the considerable amount of research showing that fathers are often absent in the lives of children growing up in FC (Farmer & Moyers, 2008;Havik, 2007;Hunt et al., 2010).Also in follow-up studies exploring contact over time, fathers absence seems to persist (Hunt et al., 2010), but see further down for a discussion of some fathers coming into contact with their children later in life.
Why are so many fathers absent?
While few Norwegian children in the general population today grow up without having contact with their fathers, the issue continues to be given a great deal of attention and is often related to negative outcomes for children (e.g.Thuen et al., 2021).From this perspective, it is reasonable to ask why so many fathers in our study did not have contact with their children.One reason can be linked to what Havik observed in her study, namely that fathers had a low involvement in the child's life also before the first placement (Havik, 1996).This was also the case in our study, as most children lived with only their mother before the FC placement.Another reason might be related to the fathers' life situation or lifestyle.Our data show that more fathers were dead while their children were young (T1), indicating that many fathers lived a more high-risk life than the mothers when the children were young.
Some would argue that little or no contact between children in care and their fathers is due to CWS failing to include fathers (e.g.Storhaug, 2015).Furthermore, from interviews with the mothers who participated in our study, we know that some mothers decided to hide the identity of the fathers from the children or encourage as little contact as possible.The main reason for this was fear of violence from the father.Importantly, this was described only by a few mothers.Nevertheless, this information, combined with CWSs deficient routines to ensure the involvement of fathers, can mean that some fathers may have little chance to initiate contact with their children during childhood.While more research is needed on possible restrictions towards fathers, the findings from our study seem to resonate with that of Hunt and colleagues (2010).Examining the cases of more than 110 children placed in KC through care proceedings in England, they found little evidence of fathers waiting to get into contact with their children.Instead, the 'Fathers (…) seemed to have just drifted out of their children's lives' (Hunt et al., 2010, p. 86).

But some children come in contact with their father later in life
Our study also shows that some children come into contact with their father later in life.Indeed, while 13% of the fathers were unknown during childhood, this percentage decreased to only 4% when the children had become young adults.This finding highlights the importance of measuring parental contact in FC over a longer time span, and perhaps the importance of exploring parental contact in adulthood.Yet, it also makes us ask what allows this contact to initiate later in the children's lives?Whether it is the fathers who take the initiative for this contact or the children themselves we know little about and is a question for future research.What we do know is that the parents of children in care die at a younger age compared to children in the general population (Backe-Hansen et al., 2014).Hence, we would probably find a noticeable difference in the possibility of contact with parents among adults who grew up in FC compared to adults in the general population where most have one or both parents alive for longer parts of adult life (Herlofson & Daatland, 2016).

Contact frequency is maintained during adolescence
Our test of the effect of time on the frequency of contact was not significant.This means that the contact pattern established in childhood seems to have continued into adulthood for most of the children in our study.Comparing this result to previous studies is difficult due to the lack of quantitative longitudinal studies exploring parental contact for children in FC.Studies examining parental contact in the general population, however, have found that children spend less time with their non-resident parent in adolescence (Lyngstad et al., 2015).Contact frequency did not decrease in adolescence among the children in our study.While we do not have data to explain this difference, qualitative interviews indicate that adolescence is a period when many children in FC express a wish to see their parents more frequently (Skoglund & Thørnblad, 2021).It is also not unlikely that foster parents open for more contact as the children get older.Moreover, parental contact for children in FC is a more regulated practice than for children in the general population, so it might be that CWS also has had a role in facilitating more parental contact during adolescence in our study population.

Mothers and daughters play important roles in maintaining contact frequency
Children have more often contact with their mother than with their father at all three points of measurement: in childhood, adolescence, and adulthood.And while there was no effect of child's gender on contact with the mother, daughters were found to have more often contact than sons with their father throughout the study.While we know little about why this is, we know from research that women are more likely to take responsibility for children and family members (Gautun, 2009).In light of this knowledge, it is reasonable to hypothesise that girls, as they get older, can take the initiative to maintain a high frequency of contact with their father.This would mean that the mothers and daughters play significant roles in determining the frequency of parental contact in FC: the mothers by being the parent maintaining contact from childhood to adulthood, and the daughters probably by initiating or continuing contact with a potentially less contactseeking father.

Limitations
One limitation of this study is that frequency of contact is measured in different ways at T1-T2 vs. T3.During this time, there has also been an extensive development related to social media which has probably had an impact on the contact frequency we measured at T3.If we had focused only on physical contact also at T3, it could be that we would have measured a decrease in contact frequency at T3, as we had hypothesised.Another limitation, which is often a concern in longitudinal studies, is the dropout we experienced from T1 to T3.Our study spanned over 16 years, and since children who changed foster home during the study period were probably less likely to be invited at T3, we can suppose that the young adults answering at T3 might not be totally representative of our original sample.The use of mixed models, however, allows to include answers from all available timepoints and therefore does not totally exclude from the analyses the children that did not participate at T3.In addition, our attrition analyses conclude that the probability to participate at T3 was not related to the contact patterns at T1, which makes us confident that the results we present are valid and free from attrition bias.
This study gives insight into stability and change in parental contact for children in FC, but not the quality of such contact.In Herlofson and Daatland (2016) research on the general population of Norway, adult children and their parents not only report frequent contact but also emotionally close relationships and exchange of helpparticularly from parents to adult children.What would young adults who grew up in FC report on such issues?In a time when contact is increasingly being facilitated, the quality of contact and the meaning of these relationships for children in care should also be taken into consideration.

Conclusion: different contact patterns for mothers and fathers
This study of contact between children in FC and their parents has given insight into different contact patterns for mothers and fathers.From a longitudinal perspective, we can say that the most significant changes evolve around children's contact with their fathers, while contact with mothers was more stable.While fathers varying contact with their children is often a topic of discussion, mothers continued contact after placement in care is rarely discussed.According to Holtan and Eriksen, mothers and fathers deprived of the right to care for their child experience different expectations -'The impossible choice [not having contact] for women is an accepted choice for men ' (2006, p. 189).Mothers continued contact deserves more attention from both social work practice and research.Notes 1. https://www.regjeringen.no/globalassets/upload/bld/lover/barnevernloven-engelsk-01-01-2010.pdf.2. The Norwegian Foster Care Association conducted a survey in 2020 where around 1000 foster parents participated.

Disclosure statement
No potential conflict of interest was reported by the authors.

Notes on contributors
Jeanette Skoglund is associate professor, Regional Centre for Child and Youth Mental Health and Child Welfare, Faculty of Health Sciences, UiT The Arctic University of Norway, Tromsø, Norway.

a
Timepoint: Timepoint 3 is the reference category.b Sex: boy is the reference category.c Type of placement 2: NKC is the reference category.

a
Timepoint: Timepoint 3 is the reference category.b Sex: boy is the reference category.c Type of placement 3: NKC is the reference category.

Table 1 .
T1 variables associated with participation at T3.

Table 2 .
Descriptive statistics showing the number of children in each category (N) and the percentage of the total sample (%).
Figure 1.Children's Possibility for contact with parents.Results given as the percentage at each timepoint.

Table 3 .
Frequency of contact with the parents for respondents having a living mother or father at the particular timepoint.

Table 4 .
Linear mixed models examining Frequency of contact with parents through time as a function of Type of placement 2 and Sex of the child.
Age in 2000 is included as a control variable.Results that are significant at p < 0.05 are bolded.

Table 5 .
Linear mixed models examining Frequency of contact with parents through time as a function of Type of placement 3 and Sex of the child.Age in 2000 is included as a control variable.Results that are significant at p < 0.05 are bolded.