Adoptive Parenting Cognitions: Acknowledgement of Differences as a Predictor of Adolescents’ Attachment to Parents

SYNOPSIS

Objective

Adoptive parents’ acknowledgement of differences is defined as the propensity to think that adoptive and nonadoptive families are different in important ways. Few studies have examined the implications of such cognitions for the parent-child bond.

Design

Structural equation modeling was utilized to examine the relation between adoptive parents’ acknowledgement of differences and adolescents’ later attachment to their parents in a sample of within-race domestic infant adoptions. Data from 189 adoptive families were drawn from two waves (middle childhood, adolescence) of the Minnesota/Texas Adoption Research Project, a longitudinal study of openness in adoption.

Results

Levels of acknowledgement of differences displayed by the adoptive mother and adoptive father during middle childhood positively predicted adopted adolescents’ feelings of attachment towards the respective parent 8 years later. This relation depended on adopted adolescents’ attitude toward adoption-related communication during middle childhood as well as the adoptive family’s level of openness during middle childhood.

Conclusions

Acknowledgement of differences in adoptive families has positive implications for the parent-child bond.

INTRODUCTION

There has been longstanding interest in adoptive parenting and attachment between adopted children and their parents because their relationship is preceded by the adopted child’s loss of relationship with his or her birth parents. To what degree do adoptive parents understand the nature of their parental role in a bionormative society that values families created through biological connection more highly than through other means, such as adoption (e.g., Baker, 2008; Witt, 2014)? Kirk (1984) provided a conceptual framework for this cognitive process in Shared Fate Theory. He posited that adoptive parenthood presents challenges not experienced by parents whose children are born to them, and adoptive parents cope through varying in acknowledging differences between adoptive and nonadoptive parenthood. Kirk (1981, 1984) provided preliminary evidence for a pathway from acknowledgement of differences to the formation of a trusting relationship between parent and child; however, a clear link between an adoptive parent’s thoughts and beliefs about adoption and parent-child attachment has yet to be empirically established.

Increased interest in parents’ mental perceptions and understanding of their children has arisen partially due to calls in the parenting field to identify stronger predictors of attachment security (van IJzendoorn, 1995). From this focus emerged efforts to conceptualize and assess parenting cognitions (Meins, 1997; Oppenheim & Koren-Karie, 2002; Shai & Fonagy, 2014;), which have been found to have important implications for the formation of secure parent-child attachment (Koren-Karie, Oppenheim, Dolev, Sher, & Etzion-Carasso, 2002; Meins, Fernyhough, Fradley, & Tuckey, 2001; Slade, Grienenberger, Bernbach, Levy, & Locker, 2005). The parenting cognitions literature provides a framework linking mental processes of parents to the parent-child bond that can be used to examine cognitive aspects of Shared Fate Theory. The purpose of the current study is to use this framework in discerning the relation between adoptive parents’ acknowledgement of differences and their adopted child’s perceptions of attachment to parents.

Kirk’s Shared Fate Theory

Kirk’s (1984) theory was developed when the prevailing attitude towards adoption was one of secrecy. Closed adoptions (in which adoptive and birth family members were not known to one another) were typical, and there was the idea in North America that adoptive families should strive to be indistinguishable from nonadoptive families (Kirk & McDaniel, 1984). Consequently, it was common for adoptive parents to downplay their child’s adopted status; some parents hid the fact of the adoption in its entirety. Extreme examples of secrecy, such as hiding the adoption from the adopted child, are currently rare, but families still differ in the degree to which they acknowledge differing from nonadoptive families (Siegel & Smith, 2012). Therefore, examining implications of such variations in acknowledgement remains important.

Shared Fate Theory posits three processes: acknowledgement of differences, empathy, and communication (Kirk, 1984). By acknowledging differences between adoptive and nonadoptive parenthood and recognizing adoption as integral to their role as parent, adoptive parents can understand that their children may have experiences and challenges unique to their adoptive status. This empathic response promotes attuned and open communication between the parent and child about adoption, as the parent is comfortable being available to his/her child for these discussions (Brodzinsky, 2005; Kirk, 1984). Openness in communication allows for the parent and child to share each other’s concerns, leading to the development of a trusting relationship and a solid foundation for secure attachment (Brodzinsky, 2005; Kirk, 1981; Kohler, Grotevant, & McRoy, 2002). In contrast, rejection of differences interferes with parents’ ability to attune to the child’s adoption-related needs. This rejection and difficulty with attunement then discourage open communication about adoption because the communication threatens the adoptive parent’s desire to ignore the facts of the adoption (Kirk, 1984). Lack of open communication strains the relationship between the parent and child because an important fact about their relationship cannot be discussed (Brodzinsky, 2005; Kohler et al., 2002). Evidence supporting this pathway has been found in Kirk (1981) as well as in an examination in which acknowledgement of differences positively and longitudinally predicted both empathy and openness in communication (Cashen, Lo, & Grotevant, 2018).

Outcome research on adopted children has largely been dedicated to the behavioral aspects of Shared Fate Theory, namely openness in communication (e.g., Kaye & Warren, 1988). A few studies examining the more cognitive construct of acknowledgement of differences have identified limited evidence that these cognitions are beneficial to both the adopted child and the parent-child bond (Brodzinsky & Reeves, 1987, as cited in Brodzinsky, 1990; Kaye, 1990; Sobol, DeLaney, & Earn, 1994). Higher levels of acknowledgement by parents are related to more problems in the adoptive family, with such findings interpreted as problems within the family leading family members to emphasize the adoption as a cause for such problems (Kaye, 1990). In addition, one study found an inverse relation between adult adoptees’ perceptions of parental acknowledgement and retrospective perceived closeness with parents at different stages in life (Sobol, DeLaney, & Earn, 1994). However, these studies conceptualized parental acknowledgement in ways that did not directly or accurately address the parent’s cognitive processes.

Such conflicting findings may be explained by hypotheses that the relation between acknowledgement of differences in parents and positive family outcomes is curvilinear (Brodzinsky, 1987). This hypothesis came from observations of extreme levels of acknowledgement in adoptive families who had sought clinical help, termed by Brodzinsky (1987) as insistence-of-differences, leading to the child not being fully integrated into the adoptive family. Brodzinsky (1987) suggested that parents may assign blame for adoption-related problems to the child’s biological differences from them, resulting in distancing or conflict between the adoptive parent and child. Overall, these discrepancies in the literature suggest a need for in-depth examination of familial outcomes of this cognitive construct.

Parenting Cognitions

When Kirk (1964) conceptualized Shared Fate Theory, there existed little emphasis on the cognitive aspects of parenthood in the parenting literature. More recently, there has been increased focus on the thought processes of parents and the implications of these processes for the child. For example, researchers have focused on the cognitions that parents have about their child’s mental state. Such a focus revisits Ainsworth’s (1969) initial conceptualization of sensitivity, which emphasizes being able to “see things from the [child’s] point of view” (p. 2). Methods of conceptualizing and assessing these cognitions currently include mind-mindedness (Meins, 1997; Meins et al., 2001), parental insightfulness (Oppenheim & Koren-Karie, 2002), and parental reflective functioning (Fonagy, Steele, Steel, Moran, & Higgitt, 1991; Slade et al., 2005). Although these conceptualizations differ in their origin and exact definition, they share similarities with social cognitive aspects of parenting such as role-taking and empathy. Specifically, they place importance in parents’ understanding and being attuned to the motivations underlying the behaviors of the child. Children are recognized as having their own, sometimes complex, thoughts that influence how they interact with the world (Meins, et al., 2001; Oppenheim & Koren-Karie, 2002). Parenting cognitions have been found to have positive implications for parent-child attachment (Meins et al., 2001; Oppenheim, Koren-Karie, Dolev, & Yirmiya, 2012; Slade et al., 2005), and they may be stronger predictors of attachment than parental sensitivity as traditionally measured (Koren-Karie et al., 2002; Meins et al., 2012).

Shared Fate Theory as Adoptive Parenting Cognitions

Significant aspects of Shared Fate Theory (Kirk, 1984) fit within the framework of general parenting cognitions. To begin, both emphasize role-taking and understanding the child’s view and mental states. This is seen most clearly in the empathy component of Shared Fate Theory, in which parents are attuned to the child’s adoption-related thoughts and experiences (Brodzinsky, 2005; Kirk, 1984). However, Shared Fate Theory posits a parent’s acknowledgement of differences must occur before adoption-related empathy may take place (Kirk, 1984), with evidence suggesting the two are closely related (Cashen, Lo, & Grotevant, 2018; Kirk, 1981). By refusing to acknowledge differences, parents may be unable to empathize with the child regarding adoption-specific challenges or understand that the child may have his or her own concerns regarding the adoption. Such lack of attunement may then interfere with the development of healthy attachment. Thus, acknowledgement of differences may be a parenting cognition specific to adoption that is predictive of attachment.

Other parallels between Shared Fate Theory and general parenting cognitions exist. For example, Shared Fate Theory describes how adoptive parents who are unable to acknowledge differences may be so distracted by their own fears that they misinterpret their adopted child’s behaviors. A parent may mistakenly think a child’s silence about adoption means that he or she is not concerned with the topic, when in fact the child does not feel as though the parent is open to such discussions. Such understanding of the motivations underlying a child’s behavior is integral to the conceptualization of general parenting cognitions (Fonagy et al., 1991; Meins, 1997; Oppenheim & Koren-Karie, 2002).

In addition, Shared Fate Theory emphasizes that through acknowledging differences, displaying empathy, and having open communication with their children about adoption, parents sacrifice their own comfort (and feelings that they are entitled to rear their adopted child) in exchange for the well-being of the child and the parent-child relationship (Kirk, 1984). Doing so may be distressing for adoptive parents, as they are forced to face their own challenging and unique role as adoptive parents as well as uncomfortable memories about the adoption process (e.g., grief over infertility, the uncertainties, feelings of powerlessness). Similar emphasis is seen in parents high in reflective functioning who are more able to empathize with a distressed child and handle their own emotional distress (Rutherford, Booth, Luyten, Bridgett, & Mayes, 2015; Rutherford, Goldberg, Luyten, Bridgett, & Mayes, 2013). Thus, in both instances, parents empathize with the child’s needs and place these needs above their own.

Finally, although acknowledgement of differences does not in itself involve attuning to the child’s internal state, there is focus in the general parenting cognitions literature on parents’ reflections on themselves. Early conceptualizations of reflective functioning emphasized parents’ reflection on their own earlier relationships as a basis for attunement to the child’s mental state (Fonagy et al., 1991). In addition, examinations of reflective functioning in adoptive families have been inclusive of parents’ views on the child’s mental states and parents’ views of themselves as a parent and their relationships with the child (e.g., León, Steele, Palacios, Romón, & Moreno, 2018; Priel, Melamed-Hass, Besser, & Kantor, 2000). Parents’ views on themselves, their own roles as parents, and their thoughts of their child’s internal states are likely intertwined.

Most research on parenting cognitions describes parents’ displaying these cognitions during the child’s infancy (e.g., Koren-Karie et al, 2002; Meins et al., 2001; Slade et al., 2005), although they continue to have important implications during childhood and adolescence (Benbassat & Priel, 2012; Oppenheim, Goldsmith, & Koren-Karie, 2004, Scopesi, Rosso, Viterbori, & Panchieri 2015). Mentalizing abilities are particularly important at the earliest developmental stage, as parents must be able to accurately read the motives of children who cannot freely communicate their own mental states. However, it is essential to examine an adoptive parent’s acknowledgement of differences beyond the child’s infancy due to the developmental nature of children’s comprehension of adoption. Although adopted children may refer to themselves as being adopted as early as preschool age, they do not fully understand the circumstances and decisions that surround adoption until middle childhood and adolescence (Brodzinsky, 2011). Adoptive parents may have to adapt their display of acknowledgement of differences to the child’s current adoption-related needs (Brodzinsky, 1987). Thus, acknowledgement of differences is particularly important when children are developing an understanding of adoption and beginning to question aspects of their own adoptions.

Acknowledgement of Differences and the Family Context

In line with the secrecy of that time, there existed little to no contact between adoptive and birth families when acknowledgement of differences was conceptualized. Often referred to as openness in adoption or family structural openness (Brodzinsky, 2005; Grotevant & McRoy, 1998), contact between adoptive and birth family members has evolved over the years toward an attitude of openness, with multiple contact arrangements currently existing (Siegel & Smith, 2012). Some adoptive families have no contact with birth family members, others have regular in-person meetings, and still others have de-identified communication that is mediated by their adoption agency (Grotevant & McRoy, 1998). Contact introduces considerations for parents’ acknowledgement of differences that were largely absent during the conceptualization of Shared Fate Theory. For example, adoptive parents may find themselves less able to reject the fact of the adoption when they and their child are reminded of it through birth family contact. In fact, children with more information about birth family members, as is often the case in more open arrangements, tend to ask their adoptive parents more adoption-related questions (Wrobel, Kohler, Grotevant, & McRoy, 1998). In addition, children with open arrangements may utilize birth family members as sources for information and adoption-related communication (Wrobel, Grotevant, Samek, & Von Korff, 2013). This circumstance contrasts with confidential arrangements, in which there is no contact and little information about birth family members available. Adoptive parents within confidential arrangements go without these reminders and subsequently have fewer threats to their ability to reject differences. As such, it is important to examine how parents’ acknowledgement of differences may interact with different openness arrangements in predicting family outcomes.

Although acknowledgement of differences involves the adoptive parents’ thought processes, its consequences involve interactions between the parent and child (Kaye & Warren, 1988; Kirk, 1984). Thus, the experiences and behaviors of adopted children play an important role in the function of acknowledgement of differences within the adoptive family. Shared Fate Theory and later work on openness in communication emphasize the parent’s ability to attune to the child’s adoption-related needs and communicate with the child appropriately, which itself is only possible through parents acknowledging the adoption as an important part of the family’s history (Brodzinsky, 2005; Kirk, 1984). Such attunement may be more necessary for adoptive parents of children who dislike communication about adoption-related topics, as these children are less likely to directly verbalize their adoption-related challenges or be open about such experiences when asked. Reasons for discomfort could be the child’s own personality traits (Brodzinsky, 2005), or there may be some existing family climate factor (such as a perception that discussion of the topic is not welcome) that dissuades the child from initiating conversations (Wrobel, Kohler, Grotevant, & McRoy, 2003). Examination of how the relation between acknowledgement of differences and attachment depends on the attitudes of the child may provide clarity to this dynamic.

The Current Study

The current study examines the relation between adoptive parents’ acknowledgement of differences and later parent-adolescent attachment. We hypothesize that acknowledgement of differences is a parenting cognition unique to adoptive families and will function similarly to general parenting cognitions in predicting attachment. Specifically, parent acknowledgement of differences during middle childhood will predict higher levels of adolescent perception of attachment to the parent during adolescence. The current study utilizes a non-clinical and low-risk sample of adoptive families (Grotevant, McRoy, Wrobel, & Ayers-Lopez, 2013). Thus, we predict and test for a linear relation between acknowledgement of differences and attachment as opposed to a curvilinear relation (Brodzinsky, 1987). Nevertheless, the sample is examined for potential cases of insistence in differences.

Family structural openness introduces additional considerations to parents’ acknowledging differences that were not present during the original conceptualization of Shared Fate Theory. In addition, children differ in how they approach the topic of adoption, and constructs described in Shared Fate Theory ultimately involve interactions between parent and child. Therefore, structural openness and the adopted child’s attitude towards adoption-related communication are examined as moderators in the relation between acknowledgement of differences and parent-adolescent attachment.

METHOD

Participants

Participants included adoptive fathers, adoptive mothers, and adopted individuals (81 females, 90 males) from 190 adoptive families who participated in the Minnesota/Texas Adoption Research Project, a longitudinal study of relationships in adoptive families. Data were taken from Wave 1 (W1: 1986–1992; middle childhood: mean age = 7.8 years, SD = 2.1 years) and Wave 2 (W2: 1996–2001; adolescence: mean age = 15.7 years, SD = 2.1 years) of the larger study. All adopted individuals were domestically adopted into same-race adoptive families as infants (mean age = 4.0 weeks, SD = 5.9 weeks) through private adoption agencies in the United States. Most adoptive families were European American (97%), and a small number identified as Latin American, African American, or Latin American and European American. Most adoptive parents identified as Protestant and middle to upper-middle class. Average number of years of education was 15.1 for adoptive mothers, SD = 2.4 years, and 16.2 for adoptive fathers, SD = 2.5 years. Detailed information on the broader study may be found in Grotevant, McRoy, Wrobel, and Ayers-Lopez (2013).

Procedure

Data collection for W1 took place through visits to the homes of the adoptive families. Sessions were approximately 3–4 hr long and included individual interviews with the members of the adoptive family as well as the administration of questionnaires. At W2, data collection similarly took place in the homes of the adoptive families and lasted approximately 4–5 hr. Data collection occurred via phone and mail when home visits were not possible for a small number of participants.

Measures

Acknowledgement of Differences

Adoptive mothers’ and adoptive fathers’ acknowledgement of differences were measured at W1 utilizing the Acknowledgement of Differences (AOD) subscale of the Kirk Adoption Questionnaire (KAQ; Kirk, 1981, as modified by McRoy, Grotevant, & Zurcher, 1988) and two items coded from interviews with the adoptive mothers and adoptive fathers. The modified KAQ is a parent self-report measure consisting of 6 items assessing aspects of the parent’s acknowledgement of differences, which are summed to create a subscale score. Examples of items include “How frequently have you recalled that at one time the child legally belonged to someone else?” and “How frequently have you wondered whether the birth mother ever thinks about the child?” A previous study examining the AOD subscale in the current sample found internal consistencies of alpha = .71 for adoptive mothers and alpha = .71 for adoptive fathers (Bohman, McRoy, & Grotevant, 1997).

At W1, adoptive mothers and adoptive fathers individually completed a semi-structured interview that included questions about their motivation for adopting, their experiences during the adoption process, and their feelings and experiences pertaining to the level of openness in the adoption. Interviews were coded globally for two aspects of adoptive parents’ acknowledgement of differences. The first was the adoptive parent’s acknowledgement of his/her family’s status as an adoptive family, rather than a nonadoptive family, and of issues that are associated with this status. The second was the adoptive parent’s interest in his/her adopted child’s pre-adoption background and history (including birth parents) outside of medical history. Both constructs were coded on 5-point scales. Coders were trained to reliability of 80% exact agreement, and two coders were assigned to code each interview. Any disagreements among coders during the coding process were resolved through discussions, and interrater reliability on these two constructs prior to resolution of discrepancies were 69% and 79% respectively.

Adolescent-Perceived Attachment

Adopted adolescents’ perception of attachment towards adoptive mother and adoptive father was measured at W2 utilizing the revised Inventory of Parent and Peer Attachment (IPPA; Armsden & Greenberg, 1987). The IPPA is an adolescent self-report measure consisting of 25 items for each parent. The IPPA’s validity has been well-established (Armsden & Greenberg, 1987; Armsden, McCauley, Greenberg, Burke, & Mitchell, 1990), and a past study utilizing the measure at W2 of the larger study found internal consistency coefficients of .96 for adolescent attachment to the adoptive mother and .97 for adolescent attachment to the adoptive father (Grant-Marsney, Grotevant, & Sayer, 2015). The IPPA consists of three subscales (Trust, Communication, and Alienation) that capture various aspects of adolescent feelings of closeness towards his/her parent. Internal consistency for the mother and father subscales of the IPPA ranged from .76 to .92.

Moderators

Adopted individuals’ attitudes towards adoption-related communication (ATAC) were measured at W1. During a semi-structured interview, adopted children were asked “Do you like to talk about your adoption?” Valid responses to this question were coded as yes, no, sometimes, or mixed. Due to the similarity in the responses, responses of sometimes and mixed were combined into a single category representing a middle point between yes and no. Adoptive family’s structural openness was coded at W1 from interviews with the adoptive mothers, adoptive fathers, and, if available, the birth mothers of the adoptive families. Integration of reports on openness from the available family members revealed four primary openness categories: confidential, time-limited mediated, ongoing mediated, and fully disclosed. Confidential adoptions (n = 62) were those in which there was no contact or transfer of information between an adoptive family and birth family beyond six months post-placement. Time-limited mediated adoptions (n = 17) were those in which non-identifying information was initially shared indirectly between an adoptive and birth family through an adoption agency caseworker; however, the sharing of information had ended by W1 data collection. Ongoing mediated adoptions (n = 52) were adoptions in which information continued to be shared indirectly through the help of a caseworker at the agency. Finally, fully disclosed adoptions (n = 59) involved contact and sharing of information that occurred directly, such as through in-person meetings, phone calls, or the exchanging of letters. Of the fully disclosed adoptions, 57 adoptive families had ongoing direct contact with the birth family, whereas two adoptive families had stopped contact with the birth family, with no plans of initiating contact in the future.

Plan of Analysis

Variables were initially screened for normality and the presence of outliers. Potential for multi-collinearity was assessed throughout analyses. Outliers were identified through examination of influence statistics (Belsley, Kuh, & Welsch, 1980), which were calculated separately for mother-adolescent and father-adolescent dyads through regressing the combined attachment score on the IPPA at W2 on the AOD scale of the KAQ at W1. When examining the model for mother-child dyads, one case was identified as being particularly influential. Influential diagnostics for the father model did not reveal any cases that were influential to a similar degree. It was decided to exclude this case from all following analyses, resulting in a total of 189 adoptive families whose data were utilized in the current study.

The primary goal of the current study was to examine the predictive effect of parent acknowledgement of differences on adolescents’ attachment towards the parent in mother-adolescent and father-adolescent dyads. We also wished to explore the potential moderating effects of adopted children’s ATAC and structural openness level in this relation. To examine these associations, a structural equation modeling (SEM) approach was utilized, with models created using maximum likelihood estimation in the Mplus 7 software (Muthén & Muthén 1998–2012). SEM allowed for the creation of latent variables for the primary constructs of interest within mother-adolescent and father-adolescent dyads: acknowledgement of differences and attachment. The SEM framework also allowed for the simultaneous estimation of mother-adolescent and father-adolescent relations as well as the modeling of dependency across dyads.

Data analyses began with a measurement model (Model 1) that established four latent factors: mother’s acknowledgment of differences, father’s acknowledgement of differences, adolescent’s attachment towards his/her mother, and adolescent’s attachment towards his/her father. The latent variables for acknowledgement of differences consisted of three indicators each: the AOD subscale of the KAQ, the interview item concerning acknowledging differences, and the interview item concerning interest in his/her adopted child’s background. To form the latent variable for adolescent’s attachment towards his/her mother and father, the IPPA was divided into its three subscales, with each subscale acting as an indicator for its appropriate latent variable. Measures used for mother-adolescent dyads were identical to those used for father-adolescent dyads. To account for this dependency in dyads, residual variances of corresponding indicators were correlated across the two dyads (e.g., adolescent’s trust towards his/her mother with adolescent’s trust towards his/her father; Kenny, Kashy, & Cook, 2006), and non-significant correlations were pared (Kline, 2011). Significant correlations between residual variances were carried over to all subsequent models.

A direct effects model (Model 2) was then created in which adolescent attachment was regressed onto acknowledgement of differences within mother-adolescent and father-adolescent dyads. Dependencies between mother’s acknowledgement of differences and father’s acknowledgement of differences and between adolescent’s attachment towards his/her mother and adolescent’s attachment towards his/her father were modeled through correlating the variances of these latent factors. These correlations were carried into all future models. Moderations by ATAC and structural openness level were assessed independently using dummy variables. Dummy variables for the respective moderators were added as predictors of adolescent’s attachment towards his/her mother and adolescent’s attachment towards his/her father, resulting in the creation of main effects models for ATAC (Model 3a) and structural openness level (Model 4a). Moderation models were then created in which dummy variables interacted with both mother’s acknowledgement of differences and father’s acknowledgement of differences through Mplus’s XWITH function (Model 3b and Model 4b). Dummy variables and interaction variables were interchanged to rotate the reference group and examine significance and size of simple slopes for all moderation conditions.

For the measurement, direct, and dummy variable main effects models, models were considered an acceptable fit if they produced multiple adequate fit measures, such as a non-significant chi-square (Bollen, 1989), a comparative fit index above .90, a standardized root mean square residual (SRMR) below .08, and a root mean square error of approximation (RMSEA) below .06 (Hu & Bentler, 1999). Improvements in model fits from dummy variable main effects models (Models 3a and 3b) to moderation models (Models 4a and 4b) were assessed through identification of lowering in Akaike information criterion (AIC) and sample-size-adjusted Bayesian information criterion (BIC; Muthén & Muthén 1998–2012).

Missing Values Analysis

Of the 189 adoptive families in the current study, 67 families had complete data on all study variables. Missing data were primarily due to participant attrition across the two waves of the study. In addition, the children in some families did not participate at W1 because the parents felt they were too young or preferred that they not participate. Results from t-test and chi-square analyses indicated that adoptive families with complete data and those with missing data did not statistically differ on child age, child gender, adoptive father level of education, and adoptive mother level of education at W1.

Missing values were addressed in Mplus through full-information maximum-likelihood estimation (FIML; Muthén & Shedden, 1999; Schafer & Graham, 2002). Proportion of available data on study variables was at least .72. This was well above the proportion of .10 needed for reliable estimation of missing values in Mplus.

RESULTS

Descriptive statistics and correlations for the raw indicator variables are presented in Table 1. Means and standard deviations for the variables by moderation conditions are presented in Table 2 and Table 3. Model fit indices for all structural equation models are presented in Table 4. Complete covariance matrices for continuous and categorical variables in structural equation models are available by request.

TABLE 1.

Zero-order Correlations, Means, and Standard Deviations for Raw Indicator Variables (input data) used in Model 1.

Wave 1 Variables	1	2	3	4	5	6
1. KAQ AOD Scale	.41**	.19*	.29**	.21*	.23*	.19*
2. AOD (Interview item 1)	.19*	.005	.38*	.26*	.20*	.25*
3. AOD (Interview item 2)	.22**	.44**	.42**	.13	.19*	.18*
Wave 2 Variables	1	2	3	4	5	6
4. IPPA – Trust	.15	.046	.29**	.62**	.76**	.78**
5. IPPA – Communication	.030	−.045	.21*	.81**	.67**	.71**
6. IPPA – Alienation (R)	.010	.15	.26**	.68**	.64**	.68**
	M(SD)
	5.77(3.06)	2.74(.55)	2.73(.68)	41.70(7.97)	31.71(8.36)	22.47(4.90)
	6.96(3.04)	2.93(.40)	2.84(.55)	42.25(7.23)	33.38(7.15)	22.88(4.96)

Note. Correlations below the diagonal are for adoptive mothers and correlations above the diagonal are for adoptive fathers.

Correlations along the diagonal are between adoptive mother values and adoptive father values. Means and standard deviations in bold are for adoptive mothers.

^*p < .05.

^**p < .01.

TABLE 2.

Means and Standard Deviations for Raw Indicator Variables by ATAC Condition

Adoptive Mothers	No	Sometimes/Mixed	Yes
KAQ AOD Scale	6.94(3.15)	6.59(2.87)	7.49(3.03)
AOD (Item 1)	2.85(.37)	3.03(.29)	2.94(.51)
AOD (Item 2)	2.78 (.57)	2.82(.51)	2.93(.55)
IPPA – Trust	42.73(6.47)	40.20(7.87)	43.60(6.02)
IPPA – Comm.	33.39(7.21)	32.53(6.95)	34.06(7.23)
IPPA – Alienation (R)	22.90(5.36)	22.57(4.22)	22.64(5.61)
Adoptive Fathers	No	Sometimes/Mixed	Yes
KAQ AOD Scale	6.00(3.81)	5.67(2.98)	6.38(2.68)
AOD (Item 1)	2.69(.57)	2.65(.59)	2.82(.50)
AOD (Item 2)	2.56(.77)	2.72(.56)	2.91(.63)
IPPA – Trust	41.26(8.23)	39.97(8.10)	43.34(7.20)
IPPA – Comm.	31.62(8.62)	30.57(8.34)	32.58(8.17)
IPPA – Alienation (R)	23.09(5.33)	20.71(5.38)	23.05(4.61)

M (SD)

TABLE 3.

Means and Standard Deviations for Raw Indicator Variables by Openness Condition

Adoptive Mothers	Confidential	TLM	OM	Disclosed
KAQ AOD Scale	5.68(2.86)	5.31(3.04)	7.50(2.52)	8.26(3.00)
AOD (Item 1)	2.89(.42)	2.81(.40)	2.86(.35)	3.05(.39)
AOD (Item 2)	2.54(.68)	2.67(.62)	2.92(.28)	3.12(.38)
IPPA – Trust	39.10(8.56)	41.64(8.66)	44.55(5.15)	43.15(6.35)
IPPA – Comm.	30.93(7.09)	32.55(8.00)	35.49(6.55)	33.93(7.06)
IPPA – Alienation (R)	20.70(5.22)	23.45(4.30)	23.31(4.79)	24.29(4.49)
Adoptive Fathers	Confidential	TLM	OM	Disclosed
KAQ AOD Scale	4.83(2.69)	3.40(2.56)	6.69(2.87)	6.60(3.11)
AOD (Item 1)	2.66(.59)	2.53(.64)	2.82(.44)	2.80(.56)
AOD (Item 2)	2.36(.82)	2.50(.63)	2.79(.54)	3.11(.36)
IPPA – Trust	41.44(7.28)	40.64(9.47)	43.05(6.03)	41.04(9.59)
IPPA – Comm.	30.44(8.62)	30.18(8.61)	32.21(8.64)	32.78(7.87)
IPPA – Alienation (R)	21.15(4.66)	22.18(4.47)	23.06(4.99)	23.23(5.02)

M (SD)

TABLE 4.

Fit Indices for Structural Equation Models

	χ2(df), p-value	CFI	SRMR	RMSEA [90% CI]	AIC	BIC
1. Measurement Model	57.97(44), p = .077	.98	.056	.041 [.000, .067]	7318.22	7321.63
2. Direct Effects Model	60.88(46), p = .070	.98	.073	.041 [.000, .067]	7317.13	7320.40
3a. Main Effects Model – Talk1	82.97(62), p = .039	.98	.070	.042 [.010, .065]	7670.62	7674.85
3b. Moderation Model – Talk2	-	-	-	-	6221.95	6211.33
4a. Main Effects Model – Openness	122.57(71), p < .001	.94	.069	.062 [.043, .080]	7766.00	7770.75
4b. Moderation Model – Openness3	-	-	-	-	7749.49	7754.68

Note. Bayesian Information Criterion (BIC) adjusted for sample size.

¹A model with only the subsample of 147 had an AIC of 6226.08 and a sample-size adjusted BIC of 6216.16.

²AIC and BIC reported for the model with “No” as the reference group. Differences across interaction models were negligible.

³AIC and BIC reported for the model with “Confidential” as the reference group. Differences across interaction models were negligible.

Measurement Model

Unstandardized factor loadings and residuals for the measurement model (Model 1) are available in Table 5, whereas factor variances, covariances, and error covariances are presented in Table 6. In creating the measurement model, residual variances of identical indicators across mother-adolescent and father-adolescent dyads were correlated, with significant and marginally significant correlations retained. Retained correlations included correlations across the AOD scales on the KAQ, the interview items concerning acknowledgement of differences, the communication subscales of the IPPA, and the alienation subscales of the IPPA. Correlations of residual variances on these indicators were retained in all future models. Examination of global fit indices indicated that the measurement model was a good fit for the data. In addition, factor loadings for the measurement model were all significant to the .01 level, providing evidence for the four-factor structure.

TABLE 5.

Maximum Likelihood Estimates of Factor Loadings and Residuals for the Measurement Model of Acknowledgement of Differences and Parent-adolescent Attachment

	Factor Loadings			Error Variances
Indicator	Unst.	SE	St.	Unst.	SE	St.
Mother AOD
KAQ AOD	1.00a	–	.31	8.36	1.02	.90
AOD Interview1	.23	.072	.54	.11	.016	.71
AOD Interview2	.47	.17	.82	.10	.045	.34
Father AOD
KAQ AOD	1.00a	–	.38	8.07	.99	.86
AOD Interview1	.23	.065	.48	.23	.031	.77
AOD Interview2	.48	.14	.83	.14	.073	.31
Attachment to Mother
IPPA – Comm.	1.00a	–	.84	15.72	2.59	.30
IPPA – Alienation	.55	.054	.68	12.57	1.64	.53
IPPA – Trust	1.17	.080	.98	2.48b	2.06	.047
Attachment to Father
IPPA – Comm.	1.00a	–	.80	24.98	3.56	.36
IPPA – Alienation	.59	.048	.83	6.95	1.10	.31
IPPA – Trust	1.11	.082	.94	7.54	2.18	.12

Note. Unst., unstandardized; St., standardized

^aNot tested for significance

^bp > .05, all other unstandardized estimates are significant to the p < .05 level.

TABLE 6.

Maximum Likelihood Estimates of Factor Variances, Covariances, and Error Covariances for the Measurement Model of Acknowledgement of Differences and Parent-adolescent Attachment

Parameter	Unstandardized	SE	Standardized
	Factor Variances and Covariances
Mother AOD	.90a	.55	1.00
Father AOD	1.36	.66	1.00
Attachment to Mother	36.61	6.05	1.00
Attachment to Father	45.32	7.89	1.00
Mother AOD with Father AOD	.65	.33	.59
Mother AOD with Attachment to Mother	1.64	.81	.29
Mother AOD with Attachment to Father	1.33a	.81	.21
Father AOD with Attachment to Mother	2.24	1.06	.32
Father AOD with Attachment to Father	2.51	1.27	.32
Attachment to Mother with Attachment to Father	28.03	5.48	.69
		Error Covariances
M. KAQ AOD with F. KAQ AOD	3.37	.77	.41
M. AOD Interview1 with F. AOD Interview1	−.027b	.016	−.17
M. IPPA – Comm. with F. IPPA – Comm.	13.16	2.34	.66
M. IPPA – Alienation with F. IPPA – Alienation	5.40	1.05	.58

Note: Error covariances were estimated in order to account for dependency across dyads.

^ap > .05

^bp < .1, all other unstandardized estimates are significant to the p < .05 level.

Direct Effects Model

Standardized coefficients for Model 2 are presented in Figure 1, whereas unstandardized coefficients and standard errors are presented in Table 7. Results of Model 2 supported the primary hypothesis. Specifically, mothers’ acknowledgement of differences significantly and positively predicted adopted adolescents’ attachment to mother, β = .28, t = 2.70. Similarly, fathers’ acknowledgement of differences predicted adolescents’ attachment to father, β = .24, t = 2.13. Examination of fit indices suggested that this model was also a good fit for the data.

Figure 1.

Structural model for Model 2. All estimates are significant to the p < .05 level unless otherwise noted. Non-significant estimates are designated “ns” whereas estimates significant to p < .1 level are designated “a”. Standardized estimates are presented, and significance is based off of more stable unstandardized estimates.

TABLE 7.

Effects for Moderation Main Effects and Interaction Models (Models 2, 3a, 3b, 4a, and 4b)

	Outcome Variable
	Attachment to Mother			Attachment to Father
Predictor Variable	Unst.	SE	St.	Unst.	SE	St.
Model 2
Mother AOD	1.70*	.80	.28	–	–	–
Father AOD	–	–	–	1.38*	.70	.24
Model 3a
Mother AOD	1.65*	.79	.27	–	–	–
Father AOD	–	–	–	1.29	.75	.22
Talk – Sometimes/Mixed	−2.72	1.55	−.20	−2.56	1.69	−.17
Talk – Yes	.012	1.41	.001	.29	1.65	.022
Model 3b
Mother AOD	3.20*	1.49	–	–	–	–
Father AOD	–	–	–	1.41	1.05	–
Talk – Sometimes/Mixed	−3.05*	1.46	–	−2.33	1,73	–
Talk – Yes	.33	1.28	–	.96	1.63	–
Mother AOD w. Talk – Sometime/Mixed	.62	1.70	–	–	–	–
Mother AOD w. Talk – Yes	−4.81*	2.18	–	–	–	–
Father AOD w. Talk – Sometimes/Mixed	–	–	–	1.49	1.61	–
Father AOD w. Talk – Yes	–	–	–	−1.88	1.61	–
Model 4a
Mother AOD	1.17	.68	.21	–	–	–
Father AOD	–	–	–	1.40*	.70	.27
Openness – TLM	2.07	1.96	.099	−.10	2.28	−.004
Openness – OM	3.80**	1.36	.28	.73	1.66	.050
Openness – Disclosed	1.77	1.48	.14	−1.03	1.76	−.072
Model 4b
Mother AOD	4.15**	1.20	–	–	–	–
Father AOD	–	–	–	2.28*	.95	–
Openness – TLM	−2.14	2.17	–	−2.35	2.44	–
Openness – OM	1.78	1.40	–	.62	1.51	–
Openness – Disclosed	−.73	1.80	–	−3.95*	1.97	–
Mother AOD w. Openness – TLM	−6.04**	1.88	–	–	–	–
Mother AOD w. Openness – OM	−4.98**	1.64	–	–	–	–
Mother AOD w. Openness – Disclosed	−2.62	1.37	–	–	–	–
Father AOD w. Openness – TLM	–	–	–	−3.39	2.01	–
Father AOD w. Openness – OM	–	–	–	−2.73	1.42	–
Father AOD w. Openness – Disclosed	–	–	–	2.08	1.58	–

Note. Unst., unstandardized; St., standardized. Standardized estimates are not available for Models 3b and 4b. The reference group for Models 3a and 3b is “No”. The reference group for Models 4a and 4b is “Confidential”. Estimates for models with other reference groups are available upon request.

^*p < .05.

^**p < .01.

Although results from Model 2 suggest an overall linear relation between parent acknowledgement of differences and child outcome, the single outlier identified in the mother-adolescent dyads provided evidence for a curvilinear relation in extreme cases (Brodzinsky, 1987). This case was characterized by having the highest level of mothers’ acknowledgement of differences and lowest level of adolescents’ attachment to mother within the sample. Examinations of interview transcripts for this case revealed evidence in line with insistence of differences. For example, the adolescent reported a history of mental health difficulties and high conflict with his parents. In addition, the adolescent’s adoptive mother appeared to attribute many of his behavioral and emotional difficulties to his adoption and genetic background.

Tests of Moderation

Coefficients and standard errors for model testing the moderating role of adopted child’s ATAC and structural openness level are presented in Table 7. For each set of analyses, dummy variables for this moderator were first added to Model 2 as predictors of adolescent attachment to mother and adolescent attachment to father (Models 3a and 4a). Although the chi-square statistics for both models were significant, all other fit indices suggested that the two main effects models were acceptable fits for the data. Interaction terms between dummy variables and the acknowledgement of differences latent variable were then added as predictors of adolescent attachment for both mother-adolescent and father-adolescent paths (Models 3b and 4b). Reductions in AIC and sample-size adjusted BIC indicated better goodness of fit from introducing these interaction terms to the models. Dummy variables were interchanged to examine simple slopes for each category of the moderating variables.

The sample size for Model 3b was reduced to 147 due to limitations in the Mplus program’s missing data capabilities. A version of Model 2 containing only the 147 cases was created, and unstandardized regression coefficients for the primary paths in this model were comparable to the ones in the model with complete data (results available on request). Thus, the decision was made to retain the 189 cases in all other models. Results from Model 3b indicated a significant and positive relation between mother acknowledgement of differences and adolescent attachment to mother when the adopted child stated that he/she did not like talking about adoption (No response), b = 3.20, SE = 1.49, p = .032, and when the adopted child either stated that he/she sometimes liked talking about adoption or gave a mixed response (Sometimes/Mixed response), b = 3.77, SE = 1.76, p = .032. For adoptive fathers, there was a marginally significant positive relation when the adopted child endorsed sometimes liking to talk about adoption or gave a mixed response, b = 2.86, SE = 1.51, p = .058. Although nonsignificant, the relation when adopted children stated they did not like talking about adoption was in the same direction in the father-adolescent path as the mother-adolescent path, b = 1.41, SE = 1.05, p = .176. Full simple slope results are presented in Table 8.

TABLE 8.

Simple Slopes Results for Interaction Models (Model 3b and 4b).

	Parent – Adolescent Dyad
	Mother – Adolescent			Father – Adolescent
Condition	Unst.	SE	p-value	Unst.	SE	p-value
Model 3b
No	3.20	1.49	.032	1.41	1.05	.176
Sometimes/Mixed	3.77	1.76	.032	2.86	1.51	.058
Yes	−1.58	1.16	.173	−.45	1.18	.701
Model 4b
Confidential	4.15	1.20	.001	2.28	.95	.017
Time-Limited Mediated	−1.89	1.12	.093	−1.12	1.73	.52
Ongoing Mediated	−.83	.94	.378	−.46	1.02	.655
Fully Disclosed	1.54	1.12	.171	4.39	1.59	.006

Note: Unst., unstandardized. Standardized estimates are not available for Models 3b and 4b.

Results from Model 4b indicated that, for families with confidential openness arrangements with no history of contact, there was a significant positive relation between acknowledgement of differences and attachment in both the adoptive mother path, b = 4.15, SE = 1.20, p = .001, and the adoptive father path, b = 2.28, SE = .95, p = .017. In addition, there was a significant and positive relation in the adoptive father path for families with fully disclosed openness arrangements, b = 4.39, SE = 1.59, p = .006. The relation in the adoptive mother path was positive but nonsignificant, b = 1.54, SE = 1.12, p = .171. Full simple slope results are presented in Table 8.

DISCUSSION

The current study conceptualized adoptive parents’ acknowledgement of differences as a parenting cognition unique to adoptive families. Acknowledgement of differences was framed as being necessary for adoptive parents to be attuned to their child’s adoption-related experiences, thereby leading to sensitive responses to their child’s adoption-related needs and promotion of attachment. Acknowledgement of differences in middle childhood positively predicted adolescent perception of attachment to the parent 8 years later in both mother-adolescent and father-adolescent dyads. The relations were moderated by the child’s attitude towards adoption-related conversation and the family’s level of structural openness.

Adoptive parenting cognitions are specifically influential in contexts where the adopted child displays mixed to negative feelings about adoption-related conversations. These children may be less likely to verbalize adoption-related difficulties that they experience. Thus, it may be particularly necessary for adoptive parents in these contexts to acknowledge and be attuned to the potential influences of the adoption. Otherwise, adoptive parents may be unaware of, or even ignore, the thoughts and feelings driving this less positive attitude towards adoption communication or adoption in general.

Adoptive parenting cognitions were predictive in the context of confidential openness arrangements, wherein secrecy about the family’s adoptive status is most possible. Findings reflect the construct’s importance when there is limited information available to the child about his/her adoption and more need for parents to be sensitive to how this information gap is affecting their child. Evidence also supported the positive function of adoptive parenting cognitions in the context of openness, seen by the relation in father-adolescent dyads in fully disclosed adoptions. Such results can be interpreted utilizing emotional distance regulation theory (Grotevant, 2009) in that parents employ their own philosophies and expectations about adoption and birth family connections in navigating birth family contact. Through acknowledgement of differences, adoptive parents may be more attuned to their child’s feelings about contact and be able to adjust levels of contact in ways that are satisfactory for the child. Appropriate adjustments to levels of contact may be most possible in fully disclosed adoptions, as identifying information of the birth family is known and communication with birth families is convenient. Such efforts by the parent in increasing satisfaction with contact may strengthen the parent-child bond (Farr, Grant-Marsney, & Grotevant, 2014).

Adoptive parenting cognitions were not found to be related to parent-adolescent attachment within mediated contact arrangements. Reasons may include heterogeneity in contact experiences within these groups across time. Previous examinations of openness in this sample revealed mediated arrangements to be the least stable from W1 to W2. In fact, roughly equal numbers of families in ongoing mediated adoptions at W1 continued with this arrangement (36%), transitioned to fully disclosed adoptions (30%), and completely stopped contact (34%) by W2 (Grotevant, McRoy, & Ayers-Lopez, 2004). In contrast, most families with confidential or fully disclosed arrangements continued with those arrangement across those 8 years. The current study may not have been able to tease apart the multiple contact experiences within these mediated contact groups, and future research should determine how adoptive parenting cognitions function in the context of significant instability in contact.

The current study is the first to examine the implications of acknowledgement of differences during childhood for the parent-child relationship during adolescence. This two time-point design provided support for temporal precedence in the relation between acknowledgement of differences and attachment. Theoretically, the study places aspects of Shared Fate Theory into the more contemporary framework of parenting cognitions. By doing so, it integrates two theories, separated by decades, that focus on the roles and thought processes of parents. It also builds on Shared Fate Theory by moving it from a sociological account of family structure to a psychological account of relationships and outcomes.

Strengths of the study also include the use of structural regression techniques that reduced measurement error and allowed for combining multiple aspects of constructs through creation of latent factors. Data from the current study included the perspectives of adoptive mothers, adoptive fathers, and the adopted individuals themselves, giving voice to multiple members of the adoptive family, allowing for the examination of separate mother-child and father-child effects, and acknowledging that any mother-child effects occur within the context of father-child relationships and vice versa. Many past studies of adoptive parent processes have focused solely on the adoptive mother when examining adoptive parents in two-parent homes (e.g., Brodzinsky, 2006; Neil, 2009). Although it is possible there may not have been significant differences in effects between mothers and fathers, parent reports were kept separate to test this idea empirically and help address the paucity of literature on adoptive fathers’ experiences.

The study utilized a national sample of adoptive families that was purposely limited to domestic infant adoptions by same-race heterosexual parents. This was originally done to account for potential confounds introduced by more contemporary and diverse forms of adoption, including transracial adoption, international adoption, adoption by sexual minority parents, and adoption of children with special needs (Jager, Putnick, & Bornstein, 2017). As a result, generalizability of the results is limited. Last, measures used in the study were primarily self-report, thus they were vulnerable to threats to construct validity.

IMPLICATIONS FOR RESEARCH, PRACTICE, AND THEORY

Findings from the current study suggest acknowledgement of differences remains important in contemporary contexts in which secrecy is less possible. Specifically, AOD positively predicted attachment in fully disclosed adoptions, and the construct of acknowledgement of differences was observed to have variability even within the more structurally open arrangements, indicating that the idea of rejecting differences did not disappear with less secrecy. Thus, AOD should continue to be examined in modern, less hidden contexts. For example, children in transracial and international placements are often more easily identifiable as adopted due to racial differences. Racial and ethnic differences within the adoptive family have implications for adoptive parenting cognitions through the addition of acknowledgement of racial and ethnic differences between family members (e.g., DeBerry, Scarr, & Weinberg, 1996; Kim, Reichwald, & Lee, 2013). Visible differences in race and ethnicity between family members suggest that rejection of racial and ethnic differences may be difficult in transracial adoptions; however, adoptive parents may still place little emphasis on the role that race and ethnicity play in their child’s life, or employ a color-blind attitude (Lee, 2003; McRoy & Zurcher, 1983).

Past findings and theory suggest a pathway from acknowledgement of differences and empathy to openness in communication and a trusting relationship. Future research should continue to validate this pathway and examine whether this pathway holds across different contexts. More research is needed to determine how adoptive parents’ understanding of their role as parents translates into behaviors that promote secure attachments between themselves and their adopted children. It is possible that no one single behavior plays the mediating role in this relation, and that instead parents’ acknowledgement of differences provides them with the ability to respond in a way that is uniquely attuned with the child’s needs and contextual factors associated with their family.

Variables in the current study were measured at single time-points; however, they are not necessarily stable over time. Structural openness may change across developmental stages (Grotevant et al., 2019). In addition, parents’ acknowledgement of differences may change in response to new openness arrangements, initiations made by the child for more contact/information, or other adoption-related experiences. Adoptees’ views on adoption-related communication may similarly shift in response to changes in family climate about openness in communication (Brodzinsky, 2005; Wrobel et al., 2003) or the child’s own developing understanding of adoption (Brodzinsky, 2011). Future research should examine how such constructs change over time, and the implications of such changes for the parent-child relationship.

Results of the current study provide support for pre-adoption psychoeducation for adoptive parents regarding philosophies about parenting. Active acknowledgement of the adoption may be key to strengthening the relationship between adoptive parents and their adopted child. Knowledge from the study may also be incorporated into clinical practice with adoptive families, particularly for cases in which the adoption is a source of tension between parent and child. Through intervention that promotes a healthy level of acknowledgement of differences, parents may learn to understand the role that the adoption is playing in familial conflicts and respond in ways that account for motivations underlying their child’s behavior.