Mediating Role of the Child’s Temperament on the Relationship Between Mother/Father’s Adoptive Parenting and Adoptee’s Social Skills: Hybrid Dyadic Analyses

Abstract Adoption provides a unique opportunity to study the concurrent effect of child’s biology-based characteristics and environmental influences on the adoptees’ development. Herein, two Hybrid Dyadic Models—an innovation in adoption research—were tested to study the mediating role of the adoptees’ negative reactivity on the relationship between mothers/fathers’ supportive (Model A) and unsupportive (Model B) parenting, and adoptees’ social skills. In a sample of 102 couples, mothers’/fathers’ reports on adoptees’ social skills and negative reactivity, and self-reports of supportive/unsupportive parenting were studied. Mothers’ and fathers’ supportive/unsupportive parenting were independently assessed, whereas the adoptees’ negative reactivity and social skills were treated as common fate variables, with both parents’ scores as indicators of a latent construct. Results were non-significant for Model A. Regarding Model B, different relationship patterns were found depending on the informant (mother/father). Higher fathers’ (not mothers’) unsupportive parenting was associated with higher negative reactivity in adoptees, which, in turn, was associated with lower adoptees’ social skills. Considering simultaneously unique and shared perspectives of both parents, this study advances adoption research strengthening the relevance of dyadic analyses when studying the adoptive family dynamics and suggests the need to consider mothers’ and fathers’ different contributions for adoptive parenting.

A child' development is the outcome of a complex interplay between individual characteristics and contextual influences.Parenting processes may be moderated or mediated by the child's individual characteristics or, conversely, they may moderate or mediate the child's individual characteristics (Reiss, Leve, & Neiderhiser, 2013).Adoption is a natural research design, allowing for the distinction of individual and context-based processes and contributes to the knowledge of the mechanisms through which children shape and are shaped by their rearing environments (Palacios, 2018;Shewark et al., 2021).However, research in this field has mostly focused on the child's heritable characteristics that influence parenting (e.g., Leve et al., 2019), the influence of child-evoked parenting on child's developmental outcomes (e.g., Harold et al., 2013), and the moderation of the child's individual characteristics in the relationship between parenting and the child's outcomes (e.g., Reiss et al., 2013;Shewark et al., 2021).
The present cross-sectional study intends to advance previous research by exploring how the child's negative reactivity (individual characteristic) is related to supportive/unsupportive adoptive parents' reactions to the child's negative emotions (context-based) and to the adoptee's social skills (child's outcome).The theoretical framework of this study is based on the differential susceptibility model (e.g., Ellis, Boyce, Belsky, Bakermans-Kranenburg, & van Ijzendoorn, 2011), and intends to explore the role of the child's temperament (negative reactivity) as mediator, not as moderator as has been usual.As such, the present literature review is organized to present evidence on the adoptees' social skills (dependent variable), the children's negative reactivity (proposed mediator) and supportive/unsupportive adoptive parenting (independent variable).
Moreover, this study is multi-informant, since it considers both mothers' and fathers' reports for all the study variables.Resorting to a Hybrid Actor-Partner and Common-Fate Model (AP-CF Hybrid Model; Ledermann & Kenny, 2012;Ledermann & Macho, 2009;Wickham & Macia, 2019), it will simultaneously consider the interdependence between the two informants and each informant individually.Using a dyadic analysis in adoption research will be an advance allowing for a better understanding of the impact of the adoptive family dynamics on the adoptees' social development.

Adopted children's social skills
Social skills are the building blocks of children's social competence and refer to socially appropriate behaviors (Gresham, Elliott, Vance, & Cook, 2011).These include cooperative and pro-social abilities, the capacity to start and maintain conversations and social connections, flexibility, and engagement in social problem-solving, as well as empathy for others' needs and feelings (Halle & Darling-Churchill, 2016), which are critical requirements for developing successful social relationships.Given the importance of social interactions in school and other social contexts, being socially skilled in middle childhood is particularly important since this period is characterized by evident challenges in the social arena requiring flexible social adaptation (Crone et al., 2020).
The earliest building blocks of social functioning lie in the quality of the child's attachment relationships to caregivers (Guyon-Harris, Humphreys, Fox, Nelson, & Zeanah, 2019).When infancy is marked by maltreatment, neglect and/or experience of collective/group care (institutionalization), as is the case for most adopted children, the overall development and, particularly, the socio-emotional one, is heavily damaged by the lack of an individualized, responsive, supportive, and stable/permanent care.Research has shown that early adversity experiences affect social competence, specifically the acquisition and performance of social skills (e.g., Cáceres, Moreno, Román, & Palacios, 2021;Soares et al., 2019).
Nevertheless, more research is needed to explore the processes that underlie the observed heterogeneity in adoptees' social competence.The present study aims to contribute to filling in this gap by studying adoptees' social skills as an outcome (dependent variable).Since social competence is context and informant-dependent, literature has highlighted the need of considering multiple informants when studying social outcomes (e.g., Huber, Plötner, In-Albon, Stadelmann, & Schmitz 2019).As such, in this study, adoptees' social skills were evaluated by adoptive mothers and fathers.

Adopted children's negative reactivity
Temperament refers to physiologically based individual differences in reactivity and self-regulation influenced by the environment (Rothbart & Bates, 2006).Temperamental behaviors are not only biologically based but can also develop over time and be explained by the quality of the interactions/ relationships with parents/caregivers (Lipscomb et al., 2011).Negative reactivity (or negative emotionality) refers to the intensity, duration, and frequency with which the child expresses negative emotions, including fear and frustration, and is considered a core component of temperament (Putnam & Stifter, 2008).According to the differential susceptibility model (Ellis et al., 2011), children with reactive temperaments may be particularly vulnerable to environment/parenting, for the worse-suffering from the impact of negative parenting practices more-as well as for the betterbenefiting from positive parenting more (Ellis et al., 2011;Tung, Noroña, Lee, Langley, & Waterman, 2018).
In adoption research, negative reactivity has often been used as a variable accounting for a genetic influence (e.g., Leve et al., 2019), and it has been mostly explored as a predictor of specific parental responses that affect the child's outcomes (e.g., Jaffari-Bimmel, Juffer, van IJzendoorn, Bakermans-Kranenburg, & Mooijaart, 2006;Shewark et al., 2021).Higher negative reactivity has been associated with worse social outcomes in adoptees (e.g., Shewark et al., 2021;Tung et al., 2018).Following this line of research, temperament was hereby considered as a mediator of the relationship between parenting and the child's outcomes.Furthermore, this study focused on middle childhood, advancing existent research on temperament that is mostly centered on infancy and early childhood (e.g., Paulussen-Hoogeboom, Stams, Hermanns, & Peetsma, 2007).Additionally, adoptees' negative reactivity was evaluated by adoptive mothers and fathers, to account for shared method variance, i.e., reducing the informantrelated bias.

Supportive/unsupportive adoptive parenting
Parents' emotion socialization plays a central role in the child's socio-emotional development (Eisenberg, Cumberland, & Spinrad, 1998).Emotion socialization broadly refers to the multitude of behaviors that parents engage in to influence children's abilities to understand, experience, and regulate their own and others' emotions (Denham, Bassett, & Wyatt, 2007).Accordingly, parents' reactions to children's negative emotions have received significant attention from scholars.A widely used framework considers that some parents' reactions to children's negative emotions (e.g., actively engaging with children's emotions and accepting their feelings), express supportive parenting, whereas minimizing, ignoring, denying, shaming, or punishing a child's emotional experience and expressions are unsupportive parenting behaviors (Eisenberg, Fabes, & Murphy, 1996).Consequently, through supportive parenting, parents may co-regulate, as well as guide children's thoughts and behaviors, during their negative emotional experiences (Castro, Halberstadt, & Garrett-Peters, 2018;Morris, Criss, Silk, & Houltberg, 2017).Contrarywise, unsupportive parenting invalidates children's negative emotions, in an explicit (e.g., "Stop crying.")or implicit way ("You're fine.";Castro et al., 2018).
Research has shown that parenting affects and is affected by the child's temperament (e.g., Kiff, Lengua, & Zalewski, 2011).Specifically, in relation to the link between supportive/unsupportive parenting and the child's negative reactivity, a significant number of studies has associated parents' supportive/unsupportive responses to the child's regulatory capacities (e.g., Samdan et al., 2020), which are directly related to negative reactivity.In a meta-analysis, Slagt, Dubas, Deković, and van Aken (2016) explored whether children vary in sensitivity to parenting depending on their temperament.Children with a more difficult temperament were reported to be more vulnerable to negative parenting but also benefited from positive parenting more, which is consistent with the theoretical framework of this study-the differential susceptibility model (Ellis et al., 2011).The support for differential susceptibility regarding negative emotionality was, however, only present when this trait was assessed during infancy (Slagt et al., 2016).The current study advanced this line of research by including middle childhood-aged adoptees.Slagt and colleagues' meta-analysis also showed that temperament traits have been essentially studied as moderators of the parenting influence (parenting-by-temperament interaction), not as mediators, as studied herein.
Moreover, the connections between supportive/unsupportive parenting and the child's social outcomes have also been well documented, showing a positive relationship between supportive parenting and social skills/ competence, and a negative relationship with unsupportive parenting (e.g., Baker, Fenning, & Crnic, 2011;Eisenberg et al., 1996).However, recent research has brought new evidence questioning if supportive parenting is always "good" for all children and in all contexts (Castro & Nelson, 2018).Mirabile, Oertwig, and Halberstadt (2018) and Nelson and Boyer (2018) presented evidence for an age-related shift in the role of supportive/ unsupportive parenting.These authors showed that for older children, unsupportive responses can play an autonomy-promotive role (Mirabile et al., 2018;Nelson & Boyer, 2018).According to Castro and Nelson (2018), parents' supportive responses to child's negative emotions may become less needed as children become older, more autonomous, and emotionally skilled.In this line of research, Castro et al. (2018) andMiller-Slough, Dunsmore, Zeman, Sanders, andPoon (2018) also showed how the context and the informant may influence the association between supportive/unsupportive parenting and children's socioemotional competence.Castro et al. (2018) tested associations between mothers' supportive reactions and children's social skills reported by mothers and teachers and showed that mothers' supportive reactions predicted more social skills in children as reported by mothers but fewer social skills when reported by teachers.Miller-Slough et al. (2018) studied mothers' and fathers' reports of their reactions to children's negative emotions and found three different patterns of families.In one of them, fathers presented higher scores of supportive and unsupportive parenting than mothers.
Despite evidence proving the relevance of supportive emotion socialization parenting, this variable has received little attention in adoption research (see Herrera, 2014;Soares et al., 2019, as exceptions).In this study, supportive and unsupportive parenting were explored as predictors of negative reactivity (child's individual characteristic-mediator) and social skills (child's social outcome).The absence of unsupportive responses to children's negative emotions does not necessarily imply supportive responses (McElwain, Halberstadt, & Volling, 2007), and this was the reason why, in the present study, two separate models, for supportive and unsupportive parenting, were tested.Moreover, in accordance with evidence suggesting that parents' emotional socialization practices operate differently with children in middle childhood compared to preschoolers (Mirabile et al., 2018;Nelson & Boyer, 2018), the current study resorted to an age-homogeneous sample (8-10 years).Finally, it is worth emphasizing that the effect of the mother's and father's supportive/unsupportive parenting were also separately considered.Recent research has shown the unique contribution of fathers to (un)supportiveness in emotion socialization (Cherry & Gerstein, 2021;Gerhardt et al., 2020;Maia & Pereira, 2021) and the dyadic nature of mothers' and fathers' responses to children's negative emotions during middle childhood, which are predicted by their own earlier responses and by their partners' responses as well (Nelson et al., 2016).

The present study
The current study intends to examine whether supportive and unsupportive parenting (independent variables) relates to the child's social skills (dependent variable) through the child's negative reactivity (mediator variable).Previous research has predominantly used mother-centered reports inducing this study to consider both mothers and fathers as informants.Specifically, the goals of this study are to analyze the: (a) relationships between supportive/unsupportive adoptive parenting and adoptees' social skills (evaluated by both parents); (b) relationships between supportive/unsupportive adoptive parenting and the adoptees' negative reactivity (evaluated by both parents); (c) relationship between the adoptees' negative reactivity and their social skills; (d) mediating role of the adoptees' reactivity on the relationships between supportive/unsupportive parenting and children's social skills; (e) differential effect of mothers' and fathers' supportive/unsupportive parenting on the adoptees' negative reactivity and social skills.
To accomplish these goals an innovative methodological procedure-AP-CF Hybrid Model (Ledermann & Kenny, 2012;Ledermann & Macho, 2009;Wickham & Macia, 2019)-was performed.This is a holistic approach that combines the Actor-Partner Interdependence Model (APIM; Kenny, Kashy, & Cook, 2006) and the Common Fate Model (CFM; Kenny & La Voie, 1985;Ledermann & Kenny, 2012), which are models of dyadic analyses.This hybrid model allows for accessing the study variables at an individual level (mother vs father) and a group level (parents of a child), which are essential in understanding the complexity of family systems.Two models (A and B) were tested differing only in the independent variable-supportive or unsupportive parenting.Figure 1 presents the overall conceptual model proposed.
Based on the reviewed evidence, it is hypothesized that: 1.The adoptees' negative reactivity mediates the relationship between supportive parenting and adoptees' social skills (indirect effect).
Higher supportive adoptive parenting will be related to lower negative reactivity in adoptees, which, in turn, will be related to more adoptees' social skills.2. The adoptees' negative reactivity mediates the relationship between unsupportive parenting and adoptees' social skills (indirect effect).Higher unsupportive adoptive parenting will be related to higher adoptees' negative reactivity, which, in turn, will be related to fewer adoptees' social skills.3. Supportive and unsupportive parenting have differential patterns of relationships with child-related variables (negative reactivity and social skills) depending on whether it is the mother's or the father's parenting.Mothers' and fathers' supportive/unsupportive parenting will have different impacts on adoptees' negative reactivity and social skills, i.e., the relationship patterns among the variables will vary according to mothers' and fathers' parenting self-reports.

Participants
Participants included 102 different-sex couples/dyads of adoptive parents-102 mothers and 102 fathers-from a larger multi-informant study on domestic adoptees' social competence, in Portugal (Soares, 2019).Mothers' ages ranged from 32 to 59 (M = 47.88 years, SD = 5.06) and fathers' ages ranged from 37 to 59 (M = 46.24,SD = 4.89).Mothers had, on average, 12.11 years of formal schooling (SD = 4.89, range 4-23) and fathers presented a mean of 11.86 years of schooling (SD = 4.85, range 0-23).No significant differences between mothers and fathers in age and schooling were found.All participants were Caucasian.
In each family, both mother and father reported on one adoptee.Out of a total of 102 adoptees, 59 (57.8%) were boys and 43 (42.2%)girls, aged 8 to 10 at data collection (M = 8.79, SD = 0.78).Children were placed in the adoptive family when aged, on average, 3.14 years (SD = 2.16, range 0.2-8.00)and, at the time of assessment, had been in their adoptive families between 1.0 and 9.4 years (M = 5.65, SD = 2.12).Before adoption, children had spent on average 14.76 months (SD = 19.02,range 0-75) within their birth family and 23.05 months (SD = 15.58,range 1-66) in care (91% in institutional care; 9% in foster families).

Child's social skills
Adoptees' social skills were assessed using the social skills scale of the parents' version of the Social Skills Improvement System-Rating Scales (SSIS-RS; Gresham & Eliott, 2008).The social skills scale includes 46 items (e.g., "Makes eye contact when talking."; "Resolves disagreements with you calmly") that evaluate social behaviors in domains of communication, cooperation, assertion, responsibility, empathy, engagement, and self-control.Each item was rated in a frequency 4-point Likert scale, ranging from 0 (never) to 3 (almost always).The total score corresponds to the mean of the items' scores.Higher scores mean more social skills.In the absence of Portuguese norms, raw scores presented by Gresham and Elliott (2008) were used to compare findings with original norms established in a U.S. Census representative sample (for the same age group and gender).The SSIS-RS was filled in, separately, by the adoptive mother and the adoptive father within each family.The SSIS-RS is widely used in research showing good reliability and validity.In this study, excellent internal consistency results were found both for mothers (α = .94)and fathers (α = .94).

Child's negative reactivity
The adoptees' negative reactivity was measured using the School-Age Temperament Inventory (SATI; McClowry, 1995).The SATI was developed as a parent-report measure to assess the child's temperament including 38 items that define four dimensions: negative reactivity, task persistence, approach/withdrawal, and activity.In this study, only the negative reactivity scale describing the intensity and frequency with which the child expresses negative affect was used.Items were rated on a 5-point Likert scale ranging from 1 = never to 5 = always (e.g., "Responds intensely to disapproval [shouts, cries, etc.]"; "Gets mad when mildly criticized").Higher scores indicated that the child rated highly on experiencing negative affect (negative reactivity).The SATI was filled in, separately, by adoptive mothers and fathers.The Cronbach's alpha for the internal consistency of negative reactivity was .89,both for mothers' and fathers' reports.

Supportive and unsupportive parenting
Supportive and unsupportive parenting was assessed using the Coping with Children's Negative Emotions Scale (CCNES; Fabes, Eisenberg, & Bernzweig, 1990).The CCNES is a 72-item scale, which presents 12 scenarios representing common emotionally evocative events that children are exposed to and in which they may experience negative emotions.For each scenario (e.g., "If my child loses some prized possession and reacts with tears, I would…"), parents are asked to indicate the degree to which they usually respond to in six ways of coping with children's negative emotions: emotion-focused ("…distract my child by talking about happy things"), problem-focused ("…help my child think of places he/she hasn't looked yet"), expressive encouragement ("…tell him/her it's ok to cry when you feel unhappy"), minimization ("…tell my child that he/she is overreacting"), punitive ("…tell him/her that's what happens when he/she is not careful) and personal distress ("…get upset with him/her for being so careless and then crying about it").Parents rate their likelihood of responding in each of the six ways on a 7-point Likert scale from 1 (very unlikely) to 7 (very likely).The supportive coping responses scale includes emotion-focused, problem-focused and expressive encouragement reactions whereas the unsupportive coping responses scale includes minimization, punitive and distress reactions.In this study, supportive parenting refers to parents' supportive reactions to the child's negative emotions and unsupportive parenting refers to parents' unsupportive reactions to the child's negative emotions.Higher mean scores are indicators of higher supportive or unsupportive parenting, respectively.Adoptive mothers and fathers, separately, filled in the CCNES.The CCNES has been widely used in parenting research and, in the present study, the supportive and unsupportive responses scales showed excellent internal reliability for mothers (α = .93and .94,respectively) and fathers (α = .93and .93).

Procedures
This study is part of a broader research, whose design and data collection procedures were approved by the Ethics Committee of the University and the Data Protection Commission (authorization 3912/2013).Close collaboration with the National Adoption Agency allowed for the selection of adoptive families according to two criteria: (a) adoptees' age ranging from 8 to 10 and (b) at least a year after the adoption placement.Whenever possible, within each family, the mother, the father, and the adopted child participated, nevertheless, in this paper only data reported by parents are analyzed.Data collection was conducted at home visits by trained interviewers, who safeguarded all ethical principles and the participants' confidentiality.Participation was voluntary and participants signed an informed consent form.No economic compensation was given for participation in the study.

Data analysis
Preliminary analyses, including basic descriptive analyses, mean differences between mothers' and fathers' reports (Student t test) and Pearson correlations to explore bivariate relationships among study variables, were performed using IBM-SPSS.The analysis of correlations was particularly relevant to protect against any multicollinearity problem.
The Hybrid AP-CF Models were tested through AMOS, modeling individual-level variables (APIM) and common fate variables (CFM) to assess whether they are significantly associated (Ledermann & Kenny, 2012;Ledermann & Macho, 2009).The adoptee's negative reactivity (mediator) and the adoptee's social skills (dependent variable) were treated as common fate variables (cf. Figure 1).Thus, the two parents' scores on these child related-variables-the mother's and the father's contributions-were indicators of latent constructs (negative reactivity and social skills, respectively; Figure 1).Structural equation modeling (SEM) was used to estimate these latent variables.As parents of the same child, the couple shares an interdependent, relational experience (Galovan, Holmes, & Proulx, 2017).Thus, the child's reactivity and social skills are used as couple-based latent variables to account for the similarity between parents when reporting on their child.The unshared variance that remains for each individual score is treated as error (Figure 1).The use of common fate variables is especially relevant when the informants' reports are highly correlated since estimation issues due to multicollinearity are avoided and measurement error is removed.
Supportive parenting (Model A-Figure 2) and unsupportive parenting (Model B-Figure 3) by the mothers and by the fathers (independent variables) were modeled as individual-level variables using independent indicators.Supportive/unsupportive parenting is not considered at the couple level of measurement and may be better suited for the APIM.This decision was also based on the pattern of correlations among variables (see Results section; Table 1).However, the mother's supportive/unsupportive parenting and the father's supportive/unsupportive parenting were  correlated to account for the influence between the mother and father (cf.Figures 1, 2 and 3).
To account for missing data, full information maximum-likelihood estimation was employed (Enders, 2010).Despite some variables having 5% to 10% missing values, the non-significant Little's MCAR (Little, 1988)   suggests that missing data in the current study are missing completely at random, χ 2 (27) = 36.65,p = .102.
To examine the links between the mothers' and fathers' supportive/ unsupportive parenting and adoptees' social skills, through the adoptees' negative reactivity rated by mothers and fathers, path analyses were conducted.To examine the presence of mediation, the significance of indirect effects (mothers' supportive parenting → negative reactivity (latent variableparents' indicator) → social skills (latent variable-parents' indicator), and fathers' supportive parenting → negative reactivity → social skills [Model 1]; mothers' unsupportive parenting → negative reactivity → social skills, and fathers' unsupportive parenting → negative reactivity → social skills [Model 2]) was tested using bootstrapping resampling (MacKinnon, Lockwood, & Williams, 2004).Specifically, 5,000 bootstrap samples were randomly drawn, and the 95% bias-corrected bootstrapped CI was calculated.A given indirect effect is significant if the respective CI does not contain zero (Shrout & Bolger, 2002).
Model fits were evaluated using the χ 2 /df, the comparative fit index (CFI) and the root mean square error of approximation (RMSEA).A good fit is defined as χ 2 /df less than 2 and acceptable when χ 2 /df is less than 3; CFI values between .90 and .95imply an acceptable fit and between .95 and 1.00 a good model fit; RMSEA values below .08 indicate acceptable model fit and below .05point to good model fit (West, Taylor, & Wu, 2012).
Findings showed that mothers' and fathers' reports on their children's social skills were positively correlated, as well as mothers' and fathers' reports of their children's negative reactivity.Correlations between mothers' and fathers' reports (intradyadic correlations) on child-related variables were large, justifying the modeling of latent constructs resorting to both parents' contributions (common fate variables).Moreover, the higher the children's negative reactivity, the lower their social skills as rated by both mothers and fathers (Table 1).
In relation to parenting, mothers' supportive/unsupportive parenting was also positively related to fathers' supportive/unsupportive parenting.Moreover, different patterns of correlations between parenting and child-related variables were observed depending on the informant parent (cf.Table 1).No significant correlations were found between mothers' supportive parenting and child-related variables (social skills and negative reactivity).However, a significant positive correlation between fathers' supportive parenting and fathers' reports on children's social skills was observed.Mothers' unsupportive parenting was positive and significantly related to mothers' reports on their children's negative reactivity.Finally, the higher the fathers' unsupportive parenting, the lower the mothers' reports on children's social skills and the higher the mothers' and fathers' reports on their children's negative reactivity (Table 1).

Hypothesis 1: Testing direct and indirect links between supportive parenting and social skills (Model A)
Figure 2 presents the Hybrid AP-CF Model testing links among supportive parenting, negative reactivity and social skills (Model A).Overall, the model was fitted to the data, presenting a good fit, χ 2 (4) = 7.27, p = .122;χ 2 /df = 1.82;CFI = .982;RMSEA = .080.The observed variables are reliable indicators of the latent constructs-negative reactivity and social skills-because they explained a variance equal to or higher than 50%.Only 2% of the variance in the children's negative reactivity was explained by supportive parenting.The total model explained 43% of the variance in adoptees' social skills (Figure 2).
Table 2 shows the unstandardized coefficients describing direct paths between studied variables.Relations between the mothers' and fathers' supportive parenting and children's social skills were both positive, but non-significant (Table 2, Figure 2).The adoptees' negative reactivity was directly and negatively associated to their social skills, suggesting that parents who reported more negative reactivity in their children attributed them fewer social skills.This is the only significant relationship observed in this model.The relationships between mothers' and fathers' supportive parenting and children's negative reactivity revealed themselves as non-significant (Table 2, Figure 2).As such, there were no significant indirect effects in this model.

Hypothesis 2: Testing direct and indirect links between unsupportive parenting and social skills (Model B)
Figure 3 presents the Hybrid AP-CF Model testing links among unsupportive parenting, negative reactivity and social skills (Model B).Overall, the model was fitted to the data, presenting an excellent fit, χ 2 (5) = 5.63; p = .344;χ 2 /df = 1.13;CFI = .997;TLI = .991;RMSEA = .035.The observed variables are reliable indicators of the latent constructs-negative reactivity and social skills-because they explained a variance equal to or higher than 50%.Eight percent of the variance on the children's negative reactivity was explained by unsupportive parenting.The total model explained 42% of the variance in adoptees' social skills (Figure 3).Table 3 shows the unstandardized coefficients of direct relations between variables.Direct relation paths between the mothers' and fathers' unsupportive parenting and the children's social skills were both non-significant.Adoptees' negative reactivity was directly and negatively associated to their social skills, suggesting that parents who reported higher scores of their children's negative reactivity also reported lower scores of their children's social skills (Table 3, Figure 3).The relationship between unsupportive parenting and negative reactivity was significant only for fathers, i.e., the higher the unsupportive parenting by fathers, the higher the children's negative reactivity.Thus, there was a significant relationship between fathers' unsupportive parenting and children's social skills, as mediated by children's negative reactivity.This indirect relationship was negative and significant because CI do not cross zero (b = −0.04,95% CI [-0.10, −0.01]).Higher fathers' unsupportive parenting was related to higher levels of adoptees' negative reactivity, which, in turn, was related to lower social skills.The results support the assumption that the association between unsupportive adoptive parenting and adoptees' social skills is totally mediated by their negative reactivity.The mediation was total because there was no significant direct effect between unsupportive parenting and social skills.

Hypothesis 3: Testing differential effects of mothers' and fathers' supportive/ unsupportive parenting
Despite being non-significant, the direction of the links between the study variables in Model A (Figure 2) was different according to whether the supportive parenting was from the mothers or fathers.Fathers' supportive parenting had no relationship (close to zero) with the adoptees' negative reactivity; however, mothers' supportive parenting had a (non-significant) negative association with adoptees' negative reactivity (i.e., there was a slight signal of a mitigating effect).
In relation to Model B (Figure 3), although the direct paths from the mothers' and fathers' unsupportive parenting to children's social skills were non-significant, they had inverse signs of association: mothers' unsupportive parenting was positively (non-significant) related to social skills and fathers' unsupportive parenting was negatively (non-significant) related to social skills (Figure 3).Furthermore, mothers' and fathers' parenting related differently to the adoptees' negative reactivity.The fathers' unsupportive parenting was positively associated with their children's negative reactivity; conversely, mothers' unsupportive parenting showed a non-significant relationship with the adoptees' negative reactivity (close to zero).The indirect pathway (unsupportive parenting-negative reactivity-social skills) was statistically significant for fathers but non-significant for mothers.

Discussion
This study examined the extent to which adoptive mothers' and fathers' supportive and unsupportive parenting were linked to adoptees' negative reactivity and social skills.Mothers' and fathers' reports were used to studying these relationships with deeper complexity, allowing for the comparison between mothers' and fathers' different influences.A Hybrid AP-CF Model was performed, which has been an under-used methodology in adoption research.
This model allowed for the simultaneous use of (a) latent variables resulting from both parents' contributions to evaluate child-related variables (common fate variables) and (b) individual-level variables (mothers' and fathers' independent indicators).The choice of these models was adequate because they presented good fit indexes and responded to the goals of the study.Findings regarding the individual contribution of each parent's report to each latent variable showed explained variances equal to or higher than 50%, thus, confirming the choice of using common fate variables.Moreover, the differential results that were found for the effect of supportive/unsupportive parenting of mothers and fathers on child-related variables also attest the option of using these variables at an individual level.These findings are cutting-edge in adoption research, which has essentially focused on the reports of mothers, showing the relevance of considering both parents for a more complete understanding of adoptees' functioning, and valuing the role of fathers in adoptive (un)supportive parenting as has been done in non-adoption research (e.g., Gerhardt et al., 2020;Maia & Pereira, 2021;Nelson et al., 2016).
Hypothesis 1 was not confirmed.The adoptees' negative reactivity did not mediate the relationship between supportive parenting and adoptees' social skills (i.e., there is no indirect effect).Directly, supportive parenting was positively, but also non-significantly, related to social skills.Hypothesis 2 was partially confirmed.The adoptees' negative reactivity mediated the effect of fathers' (not mothers') unsupportive parenting on adoptees' social skills.Higher unsupportive parenting by fathers was associated to lower social skills in adopted children, via increasing/activating the adoptees' negative reactivity.Findings showed that, in school-aged adoptees, unsupportive parenting had no direct relationship with social skills, but an indirect one mediated by children's negative reactivity.In contrast with studies in this area using temperament as a moderator (e.g., Leve et al., 2013;Reiss et al., 2013;Slagt et al., 2016), these data showed that temperament (child individual characteristic) can also be a mediator in the relationship between adoptive parenting (environment) and adoptees' outcomes.
Since the mediational hypothesis (Hypothesis 1, Model A) was not verified when supportive parenting was used as the independent variable, it can be concluded that children with a reactive temperament seem to be more vulnerable to negative/unsupportive parenting.However, these data only confirmed the impact of negative environmental influences (unsupportive parenting) on the child's vulnerable characteristics.Since the same was not true for positive environmental influences (supportive parenting), they cannot be taken as evidence of the differential susceptibility model (Ellis et al., 2011).
Overall, findings revealed that both supportive and unsupportive parenting were not directly associated to child's social outcomes.Further, results showed that mothers' unsupportive responses can be positively associated to the child's social skills.This contradicts some previous research with non-adoption samples (e.g., Baker et al., 2011;Eisenberg et al., 1996), but aligns with recent evidence, also in non-adoption research, showing that the direction of the relationship between supportive/unsupportive parenting and the child's outcomes may shift as children get older (Castro & Nelson, 2018).Similarly to the present study's findings, previous research also showed that, in middle childhood, supportive parenting can be less valuable for children, and unsupportive parenting can have an autonomy-promotive role and a positive effect on children's socio-emotional skills (e.g., Mirabile et al., 2018;Nelson & Boyer, 2018).
These findings also bring new evidence on adoptive parenting processes that have an impact on adoptees' biology-based characteristics.Model B proved that 8% of the variance in adoptees' negative reactivity was explained by adoptive parenting (unsupportiveness).As postulated by Rothbart and Bates (2006), there are biology-based individual differences that can be influenced by the environment-reactivity seems to be one of these individual differences and supportive/unsupportive parenting is one of these environmental processes.These results are particularly relevant for intervention in adoption since adoptees do not grow up with the family with whom they share genetics.Even in older children (8-10 years old), an adoptive family able to respond to the child's specific characteristics can have a positive influence on individual characteristics that are more resistant to change.
In relation to the effect of negative reactivity on social skills, findings confirmed previous evidence (e.g., Leve et al., 2019;Shewark et al., 2021;Tung et al., 2018).The relevance of considering children's individual characteristics when assessing their outcomes in middle childhood is then emphasized.
Hypothesis 3 was proved.The paths between the study variables (direct and indirect relationships) are different whether considering the influence of supportive/unsupportive parenting of mothers or fathers.This was true in relation to the direct relationship between supportive/unsupportive parenting and social skills, the direct relationship between supportive/ unsupportive parenting and negative reactivity, and the indirect relationship between supportive/unsupportive parenting and social skills.These findings add evidence to recent research valuing the fathers' role in emotional socialization process (e.g., Cherry & Gerstein, 2021;Maia & Pereira, 2021).In the present study, fathers made unique contributions to their children's socio-emotional development.Gender is an important issue in approaching how children learn about emotions, as men and women are socialized to express vs. inhibit/suppress emotions according to gendered roles defined by cultural norms that vary across communities and contexts (Gerhardt et al., 2020).The present study proved that mothers' and fathers' emotion socialization practices may be differently linked to the child's social outcomes.This is an important consideration for research, since using only one informant can bias the results, providing a limited view of a complex process.
Finally, it is noteworthy that in this study, adoptees presented, on average, normative scores of social skills, which is consistent with previous research (for a more complete research review on this topic see Barbosa-Ducharne, 2021;DeLuca et al., 2019).By explaining the role of parenting (supportive/unsupportive) and the child's temperament (negative reactivity) on the adoptees' social skills, this study links contextual and individual variables for a better understanding of the social functioning of adopted children.

Limitations
Although this cross-sectional study provides an advance in the understanding of the processes that underlie the heterogeneity among adoptees' social competence, by exploring the mediating role played by the children's temperament in the relationship between the parents' emotion socialization and the adoptees' social skills, it presents some limitations.First, due to the cross-sectional nature of the study, the direction of the relationships was theoretically stated and statistically tested.SEM was used to overcome this shortcoming.However, further research with a longitudinal design should confirm these findings.
Second, the study variables were operationalized in specific measures, requiring some caution in the reading of the findings.Indeed, in this article, supportive and unsupportive parenting refer only to parents' reactions to children's negative emotions, and for the child's temperament, only the negative reactivity trait was considered.In the same line, in this study, only the social skill component of social competence was studied.Children's behavior problems, which compete with socially adjusted performance, were not included in the models that were analyzed.Future research can further explore the role of temperament traits in the relationship between parenting and children's behavior providing a more complete approach of children's social competence in relation to individual characteristics.Future research can also include the adoptees' reports/ voices and study the congruence between parents and children on these studied variables.
Moreover, the explored models did not include covariate variables.Indeed, since no other study variable presented significant relationships with the dependent one-adoptees' social skills-there was no need to add covariates to the models.However, future research can explore other covariates, and replicate the models searching for differences according to the child's gender, and/or controlling for the effect of the child's behavior problems, for example.
Finally, the sample of the study is a homogeneous one of domestic adoptees and two-parent families only.In fact, the aim of the study was to capture the complexity of both the mother's and the father's influences on the child's developmental outcomes, which required selecting two-parent families out of the sample of a larger research project.It is unknown if parents' supportive and unsupportive influences remain the same in single-mother or single-father families.

Conclusions
Despite these limitations, this study makes a major contribution to the understanding of adoptive family dynamics and their influence on the adoptees' developmental outcomes.Firstly, the study's findings allowed for identifying that adoptive mothers and fathers play different roles in the emotion socialization process, which led to different influences on adoptees' social outcomes.Unsupportive parenting by adoptive fathers (not by mothers) activates the negative reactivity of adopted children, which in turn reduces their ability to show adequate social behaviors.As such, secondly, this study showed that adoptees' social skills are affected both by the children's negative reactivity (directly) and by the fathers' minimization, punishment and distress when faced with their children's negative emotions (indirectly).This is an important advance in understanding how parenting impacts children since the effects are both direct and indirect.Resorting to a within-family approach to emotion socialization using an innovative methodological procedure, the current research contributes with new evidence to a better understanding of the processes underlying adopted children's social competence, which is still a field in need of further research.

Implications for practice
Overall, these findings bring evidence on parenting processes that influence, i.e., activate or mitigate, the child's individual characteristics, showing that there are biology-based individual differences, such as negative reactivity, that can be influenced by the environment, such supportive/unsupportive parenting.This evidence has important implications for professional practice with adoptive families, highlighting the relevance of considering parents' emotion socialization as a topic in both the preparation of prospective adopters, and the provision of post-adoption services.As the child's emotional learning needs change with age, aid in adoptive parents' supportive/unsupportive responses to their children's negative emotions becomes critical.Parents should adjust their socialization practices as their child grows and according to his/her socialization challenges, needs and expectations.Further, findings also pointed out the importance of promoting parenting skills able to "counteract/mitigate" reactive temperaments that negatively impact the child's outcomes.Finally, this study showed how crucial the adoptive fathers' involvement is in the process of parenting adopted children.

Figure 2 .
Figure 2. Hybrid Ap-CF: Model A-supportive parenting predicting negative reactivity and social skills.

Table 1 .
descriptive statistics and bivariate correlations for study variables.

Table 2 .
unstandardized estimates describing direct paths in Model A.
Note.Significant paths are highlighted in bold.no significant indirect effects were observed.

Table 3 .
unstandardized estimates describing direct paths in Model B.
Note.Significant paths are highlighted in bold.