Attachment and Autonomy in Adolescence 1 Adolescents’ and Mothers’ Models of Attachment as Predictors of Developing Autonomy and Relatedness in Observed Family Interactions

This study examined links between adolescent and maternal attachment models and the adolescent developmental task of attaining autonomy while maintaining relatedness with parents as observed in family interactions of 88 adolescents who were seen at ages 16 and 18. Both adolescents’ and mothers’ attachment models combined to predict interactive behaviors around autonomy and relatedness both cross-sectionally and longitudinally. Attachment security predicted higher levels of autonomy and relatedness-promoting behaviors, whereas attachment preoccupation predicted theoretically meaningful combinations of both positive and negative family behaviors. Mothers’ secure attachment models predicted both adolescent and maternal relatedness-promoting behaviors over time. Increases in adolescents’ hostile and critical behaviors (one marker of undermining relatedness) were predicted by mothers’ preoccupied attachment models. The implications of connections between attachment models, the development of family interactions over time, and broader theories of adolescent and family development are discussed. Attachment and Autonomy in Adolescence 3 Adolescents’ and Mothers’ Models of Attachment as Predictors of Developing Autonomy and Relatedness in Observed Family Interactions Adolescence has traditionally been defined as the transition from childhood to adulthood. While proceeding through this period, adolescents are faced with numerous tasks that they must master before becoming adults. The developmental task of attaining autonomy, becoming independent in one’s thoughts and opinions as well as actions, has long been thought of as one of the central processes of adolescence (Hill & Holmbeck, 1986). Contrary to traditional beliefs, growing evidence indicates that this process is most easily navigated in the context of a close relationship with parents rather than at the expense of this relationship (Allen, Aber, & Leadbeater, 1990; Allen & Hauser, 1996; Allen, Hauser, Bell, & O’Connor, 1994a; Allen, Hauser, Eickholt, Bell, & O’Connor, 1994b; Best, Hauser, & Allen, 1997; Connell, Halpern-Felher, Clifford, Crichlow, & Usinger, 1995; Grotevant & Cooper, 1985; Taub, 1997). In order to assess how well adolescents are able to strive for autonomy while maintaining a positive relationship, much of the relevant research has investigated parent-child interactions and the behaviors exhibited by each family member during these interactions (e.g. Alexander, 1973; Allen et al., 1994a; Allen & Hauser, 1996; Becker-Stoll & Fremmer-Bombik, 1997; Kobak, Cole, Ferenz-Gillies, Fleming, & Gamble, 1993; Kobak, Ferenz-Gillies, Everhart, & Seabrook, 1994; Kobak, Sudler, & Gamble, 1991; Steinberg, 1981; Steinberg & Hill, 1978). The degree to which adolescents attain autonomy during a family discussion is often measured through the quality of their reasoning and their level of confidence. The level of closeness or relatedness within the relationship is similarly identified Attachment and Autonomy in Adolescence 4 from individual behaviors during a family disagreement, such as questioning intended to further discussion and express empathy, validation of others’ ideas and opinions, and engagement in the conversation. Adolescents’ abilities to establish autonomy and relatedness as previously defined have been linked to numerous positive outcomes. For instance, research has shown that expressing autonomy and relatedness is related to positive self-esteem, ego development, ego resiliency, friendship competence and educational attainment (Allen et al., 1994a; Best et al., 1997; Connell et al., 1995; Freitag, Belsky, Grossmann, Grossmann, & Scheuerer-Englisch, 1996). Conversely, when adolescents’ autonomy is undermined within the parent-child dyad during a disagreement, it has been found that teenagers are more likely to experience internalizing behaviors such as depressed affect and anxiety. They are also more likely to become enmeshed in and confused by attachment relationships later in life (Allen et al., 1990; Allen et al., 1994b; Allen & Hauser, 1996; Fauber, Forehand, Thomas, & Wierson, 1990; Gjerde & Block, 1991; Kobak et al., 1991). In addition, a lack of relatedness between adolescents and their parents during disagreements has been linked to externalizing behaviors such as delinquency and dropping out of high school (Allen et al., 1990; Allen et al., 1994b; Allen & Hauser, 1992; Connell et al., 1995). Although much research has shown that the struggle for autonomy is a key task of adolescence and, more specifically, that maintaining relatedness with parents during this process is important and necessary, surprisingly little research has investigated the development of individual differences in this achievement. The few studies that have attempted to present possible models for this process have Attachment and Autonomy in Adolescence 5 focused on attachment theory as one possible source of influence (Allen & Land, in press; Collins & Repinski, 1994; Freitag et al., 1996; Kenny, 1994; Schultheiss & Blustein, 1994). Attachment theory focuses on the attachment relationship between children and their caregivers, and the experiences associated with those relationships (Bowlby, 1969/1982). Initially used to describe infants’ connections to their mothers and found to predict functioning five to ten years into the future (Ainsworth, Blehar, Waters, & Wall, 1978), attachment theory was later applied to adults’ strategies for processing affect and memories relating to attachment experiences described in the Adult Attachment Interview (AAI) (Main & Goldwyn, in press). Research has also shown that attachment theory can similarly be applied to describe adolescents’ memories and representations of attachment experiences (Allen & Hauser, 1996; Allen, Moore, Kuperminc, & Bell, 1998; Allen & Land, in press). Secure attachment representations in both adolescence and adulthood are characterized by accurate recollections of affect and memories regarding attachment experiences. Insecure attachment representations can be either preoccupied or dismissing in nature and are characterized by memories of attachment experiences that are inconsistent with associated affect. Individuals with preoccupied states of mind regarding attachment tend to be overly caught up in particular attachment relationships, are often confused about their attachment experiences and relationships, and seem to be unable to move beyond these experiences. Individuals who are dismissing of attachment appear to cut off and/or completely devalue attachment relationships and experiences (Main & Goldwyn, in press). The little research that currently exists on the adolescent task of attaining autonomy and relatedness looked solely at either the influences of adolescents’ states of mind regarding the attachment Attachment and Autonomy in Adolescence 6 relationship or parents’ attachment organizations. In addition, the majority of this research observed adolescents and their parents at only one point in time, thus making the research, thus far, incomplete. Because relationships involve more than one person, it can be inferred that the task of maintaining closeness within a relationship is influenced by aspects of the family as a system rather than by any one individual (Hill & Holmbeck, 1986). Previous research has, thereby, ignored the possibilities that a) adolescent and parental states of mind with regards to attachment may differ b) each individual’s state of mind regarding attachment may influence the relationship separately, as well as jointly, and c) children and their parents have reciprocal effects on each other (Bell, 1979). In addition, the nature of the parent-child relationship changes through adolescence, with adolescents turning to their parents for different types of support as they age (Allen & Land, in press; Grotevant & Cooper, 1985; Taub, 1997). Most previous studies on the development of individual differences in attaining autonomy and relatedness, however, observed adolescents and their parents at only one point in time, ignoring the fact that adolescence is inherently a time of change and transition, with the parent-child relationship changing as well. Based on attachment theory, much research has explored the link between parental attachment strategies and behaviors exhibited during parent-child interactions during infancy. Such research has shown that parents holding secure attachment representations allowed their infant children more autonomy while maintaining a sense of connectedness with their infants, whereas parents adhering to more insecure mental models of attachment had more trouble promoting a balance of autonomy and relatedness. For example, Glachan and Murray (1997) found that mothers’ attachment representations Attachment and Autonomy in Adolescence 7 were directly related to their current relationship with their infant. More specifically, parents with more insecure attachment models had infants who displayed less warmth in an interaction than children whose parents with more secure attachment organizations (Cohn, Cowan, Cowan, & Pearson, 1992). Similarly, research has shown that mothers who held insecure attachment models behaved less supportively toward their twoto four-year-old children than those with secure attachment representations during an interaction task (Rhodes, Simpson, & Blakely, 1995). Crandell, Fitzgerald, and Whipple (1997) observed that during interactions, mothers with secure attachment representations encouraged child autonomy more than mothers with more insecure attachment representations. Parentchild dyads in which the mothers held more secure attachment organizations also related to each other more warmly than those with mothers who had more insecure attachment representations. However, studies have been inconsistent in demonstrating that behaviors exhibited in the strange situation at infancy parallel behaviors of adolescence (Weinfield, 1996; Zimmermann, Fremmer-Bombik, Spangler, & Grossmann, 1995), so conclusions drawn from these studies are not necessarily applicable to adolescence. Similar findings were observed in interactions between mothers and ten-year-old children (Freitag et al., 1996). In their study, the authors combined both the mothers’ and children’s attachment representations (assessed during infancy) into a composite attachment security score. The researchers observed that the level of security was related to children’s portrayal of “connectedness” and “individuality” assessed in a family interaction task, constructs similar to relatedness and autonomy. Freitag et al. expanded the previous research on the development of autonomy and relatedness by Attachment and Autonomy in Adolescence 8 investigating the attachment as a possible predictor of children’s behaviors within the parent-child dyad. However, attachment models have been shown to change between infancy and adolescence (BeckerStoll & Fremmer-Bombik, 1997), so these findings can not necessarily be applied to explain similar processes in adolescence. In adolescence, research on attachment models and family interactions is much more limited. Kobak et al. (1991) studied the process of attaining autonomy while maintaining relatedness in adolescence by investigating relations between adolescents’ own attachment representations and behaviors exhibited during family interactions. Results indicated that adolescents’ insecure/preoccupied attachment models were associated with “maternal dominance” and adolescents’ “dysfunctional anger” during mother-adolescent problem-solving tasks. In this study, “maternal dominance” referred to the exhibition of a mother’s autonomy and the inhibition of an adolescent’s autonomy; and “dysfunctional anger” refers to the expression of negative relatedness, specifically hostile or critical behaviors. Therefore, the researchers concluded that adolescents’ attachment models were linked to both the mother’s and adolescent’s behaviors in the context of family disagreements. Kobak et al. (1993) further investigated the influence of adolescent attachment on adolescents’ current relationships with their mothers, demonstrating that gender played a role in these influences. They found that male adolescents with secure attachment representations interacted with their mothers in a manner that involved less “dysfunctional anger” and less “avoidance” (two examples of relatednessinhibiting behaviors), than males with insecure models of attachment, whereas female adolescents who held secure attachment representations exhibited only less “dysfunctional anger” than their insecurely Attachment and Autonomy in Adolescence 9 attached counterparts. Males with insecure/dismissing attachment models exhibited more “dysfunctional anger” in their interactions while interactions involving females who held insecure/dismissing attachment representations were characterized by “maternal dominance,” behavior inhibiting the adolescent’s autonomy. Results also showed that males tended to be more insecure/dismissing than girls. These findings further support the notion that attachment representations are related to family behaviors. Once again, the authors neglected to address the possibility that family members’ attachment models may contribute independently and in conjunction to influence the parent-child relationship. Kobak et al. (1994) investigated the relation between maternal attachment representations, as opposed to those of adolescents, and family behaviors. The authors observed that mothers’ preoccupation with attachment was related to adolescents’ inability to exhibit autonomy in the course of a discussion, but only with older adolescents (approximately 17 years old). No significant results indicated any relationship between mothers’ secure attachment models with behaviors exhibited by their adolescent children nor between any attachment representations and adolescents’ relatedness. These finding further support the idea that attachment models predict behaviors around autonomy and relatedness within the parent-child dyad, adding the notion that mothers’ attachment models (especially preoccupied attachment models) have predictive value, as well. Adolescents’ attachment representations were again examined by Allen and Hauser (1996). Unlike previous research, Allen and Hauser looked at interactions between adolescents and their fathers as well as with their mothers. Disagreements were observed when the adolescents were fourteen years old. Their attachment representations were assessed eleven years later using the Adult Attachment Attachment and Autonomy in Adolescence 10 Interview. Results showed that mothers’ promotions of autonomy during a disagreement were related to adolescents’ secure attachment representations eleven years later. The authors also found that adolescents who were highly autonomous while maintaining relatedness with their fathers were later more likely to be securely than insecurely attached. Also, adolescents who inhibited their fathers’ autonomy through overpersonalizing statements were most often preoccupied with attachment (unable to move beyond early attachment relationships and experiences) eleven years later. Although this study did examine some development through time, the variables of attachment and autonomy and relatedness were not measured at both times, so true change over time was undetectable. In addition and unlike previous research (Freitag et al.), this study investigated levels of autonomy and relatedness as seen in family behaviors as predictors of attachment models. Therefore, conclusions regarding the development of the adolescent task of attaining autonomy while maintaining relatedness can not necessarily be drawn from these findings. Becker-Stoll and Fremmer-Bombik (1997) once again examined cross-sectional relations between adolescent attachment models and family members’ behaviors in the course of disagreements with their parents. Like Allen and Hauser (1996) before them, the authors found that adolescents with secure attachment representations were most likely to exhibit and promote autonomy and relatedness when interacting with their mothers. Results also showed that insecure/dismissing adolescents exhibited and promoted autonomy less than their more secure counterparts and promoted relatedness less than both secure and insecure/preoccupied adolescents. Unlike previous studies on attachment and adolescent behaviors, Becker-Stoll and Fremmer-Bombik studied a sample with a wide variety of Attachment and Autonomy in Adolescence 11 socioeconomic backgrounds, demonstrating that the link between adolescent attachment representations and family behaviors appear to be consistent for various types of families. The relatively little research that exists on adolescence, and specifically that which has investigated behaviors indicative of the struggles for autonomy and relatedness, has identified attachment representation as an important possible predictor of adolescent behavior within the parent-child dyad. Studies have suggested that individual secure attachment representations predict more autonomyand relatedness-promoting behaviors, while individual preoccupied attachment models predict more autonomyand relatedness-undermining behaviors. However, no study to date has examined the independent and combined influences of both parental and adolescent attachment representations on the parent-child relationship. Further research is needed to investigate whether each person’s model of attachment contributes independently to the family interaction or interaction behaviors are primarily determined by one or the other’s model of attachment. Another question that remains unanswered is how the process of attaining autonomy and relatedness develops over time in families whose members have different attachment models. The previous research demonstrated that attachment models crosssectionally predict displays of autonomy and relatedness in family interactions. But do these predictions change over time? The rapidly changing nature of the parent-child relationship during adolescence (Allen et al., 1998) suggests that the links between attachment models and autonomy and relatedness may change, as well. Yet this has never been empirically investigated. This study further investigates individual and family differences in the developmental task of striving for autonomy while maintaining relatedness. Observations of mother-adolescent interactions and Attachment and Autonomy in Adolescence 12 individual interviews will be used to study the relationship between adolescents’ and mothers’ internal representations of attachment and family members’ behaviors within the relationship. Unlike previous research, the present study will assess the separate and combined influences adolescent and maternal attachment representations in order to better explain these processes. Incorporating the notion that adolescence is a transitional period involving changes within the parent-child relationship, the present study will also go beyond past research by observing changes in family interactions at two different points in adolescent development. The current study will also expand existing research by exploring which specific autonomy and relatedness behaviors are most linked to attachment representations. In this study we plan to address the following specific questions and hypotheses: 1) How do attachment representations manifest themselves in family interactions? It is expected that individuals with more secure attachment states of mind will promote autonomy while maintaining relatedness, while individuals who have more preoccupied attachment models will have difficulty maintaining autonomy and relatedness. 2) Do adolescent and maternal attachment representations combine to influence autonomy and relatedness? It is expected that families in which both the adolescent and mother hold more secure attachment representations will be those most likely to promote adolescent autonomy and relatedness, while families in which both members have more preoccupied attachment models will be those most likely to undermine autonomy and relatedness. 3) How do the relative influences of attachment representations change over time? It is expected that families with more secure models of attachment will increase the promotion of autonomy and relatedness over time more than their preoccupied counterparts. This study examines connections between attachment models and family interaction Attachment and Autonomy in Adolescence 13 behaviors in an ethnically and socioeconomically diverse sample of moderately at-risk adolescents. The sample was selected to allow a maximally meaningful range of family structures and types at various levels of psychosocial functioning, including substantial numbers of families and adolescents functioning both adequately and poorly. Method Participants Eighty-eight ninth and tenth graders (45 male, 43 female) and their mothers participated in this study. Upon entering the study, adolescents ranged from age 14 to 18.75 years with a mean of 15.9 years (SD = 0.9). At the second wave of research, approximately two years later, the mean age of adolescents was 18.1 years (SD = 1.0) with a range from 15.9 to 22 years. Sixty-one adolescents identified themselves as European-American (69.3%), 29 as African-American (29.6%), and one as a member of another group (1.1%). Forty-three percent of adolescents were living with both biological parents when they entered the study. The median family income was $25,000 with a range from less than $5,000 to greater than $60,000. Parents’ median education level was some college or technical school training beyond high school (range was from less than an eighth-grade education to completion of an advanced degree). At the second wave of data collection, 59.8% of adolescents were still in high school, 14.9% had dropped out, 18.4% had already graduated, 5.7% were attending college, and 1.1% had obtained a general equivalency diploma (GED). Participants were recruited through public school systems serving rural, suburban, and moderately urban areas. Adolescents were selected for inclusion in this study on the presence of at Attachment and Autonomy in Adolescence 14 least one of four possible academic risk factors: 1) failing a single course for a single marking period, 2) any lifetime history of grade retention, 3) 10 or more absences in one marking period, or 4) any history of school suspension. These broad selection criteria were used to include a wide range of adolescents who could be identified from academic records as having the potential for future academic and social difficulties. Included in the sample are both adolescents who had already experienced serious difficulties and others who were performing adequately with only occasional minor problems. As intended, these criteria identified approximately one-half of all ninthand tenth-grade students as eligible for the study. Procedure After adolescents were identified as meeting the criteria of the study, letters explaining the study were sent to the families of each potential participant. Interested families sent back post cards containing information about how and when to contact them by phone. If both the adolescent and the parent(s) agreed to participate, they were scheduled to come in for two, three-hour sessions. Approximately 67% of the families contacted by phone agreed to participate in the study. Families were paid a total of $105.00 for their participation. At each session, active informed consent was obtained from both parents and adolescents. Adolescents and their parents were interviewed separately, and in the introduction and throughout the sessions, confidentiality was assured to all family members for all data collected. Two years following the initial session, the families were again contacted by phone and invited to return for two more sessions. Again, families were paid $105.00 for their participation, active informed consent was obtained, and confidentiality was assured. All data in the study were covered under a Attachment and Autonomy in Adolescence 15 Department of Health and Human services Confidentiality Certificate which protects data against subpoena by federal, state, or local courts and other agencies. Measures Adult Attachment Interview and Q-set (George, Kaplan, & Main, 1996; Kobak, Cole, Ferenz-Gillies, Fleming, & Gamble, 1993). Researchers administered this structured interview to explore individuals’ evaluations and descriptions of their childhood relationships with parents. Participants were probed for abstract terms as well as specific supporting memories. For instance, interviewers asked participants to list five words describing their early attachment relationships and then to describe specific instances that corresponded to each word. Other questions addressed experiences of upset, separation, loss, trauma, and rejection. Finally, interviewers asked participants to provide more integrative descriptions of changes in the relationship and the current state of the relationships. The interview consisted of 18 questions and lasted one hour on average. Questions from the original adult version of this interview were slightly adapted to better accommodate an adolescent population (Ward & Carlson, 1995). Interviews were audiotaped and then transcribed for coding. The AAI Q-Set (Kobak et al., 1993). This Q-Set was designed to resemble the Adult Attachment Interview Classification System (Main & Goldwyn, in press) while assessing continuous qualities of attachment organization. [The data produced by this system can be reduced via an algorithm to classifications that largely agree with three-category ratings from the AAI Classification System (Borman-Spurrell, Allen, Hauser, Carter, & Cole-Detke, 1995; Kobak, al., 1993)]. Each coder using the Q-Set method reads an AAI transcript and provides a Q-sort description by using a forced Attachment and Autonomy in Adolescence 16 distribution and assigning 100 items into nine categories ranging from most to least characteristic of the interview. All interviews were blindly rated by at least two coders with extensive training in both the Qsort and Main Adult Attachment Interview Classification System. The Q-sets were then compared with dimensional prototype sorts for: secure versus anxious interview strategies, reflecting an overall cohesion of discourse, an accurate integration of episodic and semantic attachment memories, and a clear objective evaluation of attachment; preoccupied strategies, reflecting either a rambling and unfocused nature of dialogue on attachment experiences or an angry preoccupation with attachment figures; dismissing strategies, reflecting an inability or unwillingness to discuss memories and issues of attachment, an inaccurate idealization of attachment figures and the relationship, and/or a lack of evidence for valuing attachment, in general; and deactivating versus hyperactivating strategies, representing the overall balance of dismissing and preoccupied styles. Kobak et al. (1993) validated the use of these dimensions and stated that they accurately capture the constructs of the AAI Classification System. Each participant’s scale score consisted of the correlation of the 100 Q-sort items with each attachment dimension (ranging from -1.00 to 1.00). The SpearmanBrown reliabilities for the final scale scores were .84 for security, .89 for dismissal of attachment, .82 for preoccupation, and .91 for the hyperactivating versus deactivating scales. Although this system was designed to yield continuous measures of attachment organizations rather than replicate classifications from the Main and Goldwyn (in press) system, the current study reduced the scale scores to classifications by using the largest Q-scale score above .20 as the primary classification (Kobak et al., 1993). When scores were compared to a subsample (N=76) of the adolescent AAI’s classified by an Attachment and Autonomy in Adolescence 17 independent coder with well-established reliability in classifying AAI’s (U. Wartner), 74% received identical codes (kappa = .56, p <.0001), and 84% of scores matched in terms of security versus insecurity (kappa = .68). Family Interaction Task. Adolescents and their mothers participated in a revealed differences task in which they discussed a issue of disagreement. Typical topics included money, grades, household rules, friends, brothers and sisters, communication, plans for the future, alcohol and drugs, religion, and dating. These interactions were videotaped and transcribed for coding. Autonomy and Relatedness Coding System (Allen, Hauser, Bell, Boykin, & Tate, 1995). Coders used both the videotapes and transcripts of mother-adolescent interactions to determine the extent to which autonomy and relatedness were exhibited and/or undermined throughout the course of the discussion. Coders followed concrete behavioral guidelines provided by the Autonomy and Relatedness Coding System (Allen et al., 1995) to rate both mothers’ and adolescents’ individual statements on one or more of 10 subscales with scores ranging from zero to four. Subscales were grouped into four categories: 1) promoting autonomy, including subscales of stating reasons and displaying confidence; 2) undermining autonomy, including subscales of recanting one’s own position, overpersonalizing the discussion, and pressuring the other to agree; 3) promoting relatedness, including subscales of asking questions, validating the other person, and actively engaging in discussion; and 4) undermining relatedness, including subscales of interrupting the other person and displaying hostility during the interaction. The global scale of promoting autonomy is designed to capture the degree to which an individual can think and act independently while allowing and encouraging the other member of Attachment and Autonomy in Adolescence 18 the dyad to attain autonomy as well. In contrast, the undermining autonomy scale measures the degree to which each individual attempts to control the conversation, thereby not allowing the other person to strive for autonomy. The relatedness scales are similarly designed to capture how well each family member maintains the closeness of the relationship during a disagreement. Specific guidelines for each code and example statements are presented in Appendix A. Each interaction was coded by at least two trained coders, and inter-rater reliability for these scales was calculated using Spearman-Brown correlations. Reliability coefficients for each scale were .86 for adolescents exhibiting autonomy, .77 for adolescents undermining autonomy, .82 for mothers exhibiting autonomy, .71 for mothers undermining autonomy, .82 for adolescents exhibiting relatedness, .80 for adolescents undermining relatedness, .84 for mothers exhibiting relatedness, and .80 for mothers undermining relatedness. Past research has found this coding system to be a reliable predictor of both family and adolescent functioning (Allen et al., 1994a; Allen et al., 1994b; Allen & Hauser, 1996; Becker-Stoll & Fremmer-Bombik, 1997). Attachment and Autonomy in Adolescence 19 Results Preliminary Analyses Sample means. Means and standard deviations of measures of attachment are presented in Table 1. Means and standard deviations of the global autonomy and relatedness scales are presented in Table 2, and means and standard deviations of specific autonomy and relatedness subscales are presented in Table 3. Also presented in Tables 2 and 3 are the t-test scores for differences between mean scores on all autonomy and relatedness scales when adolescents were 18 and 16 years old. Analyses revealed that both adolescents (t(88) = 3.71, p < .001) and their mothers (t(88) = 2.92, p < .01) undermined relatedness more when adolescents were 18 than 16 years old. Results also showed that mothers increased their undermining of adolescents’ autonomy (t(88) = 4.98, p < .001) over the two-year period. No changes in adolescents’ promotion of autonomy, undermining of autonomy, or promotion of relatedness were observed over the two-year period. In addition, mothers promoted autonomy ( t(88) = 4.98, p < .001) and relatedness (t(88) = -4.42, p < .001) less as adolescents got older (see Table 2). Similar patterns of change emerged when each global autonomy and relatedness scale was broken down into its specific subscales (see Table 3). Table 1 Means and Standard Deviations of Attachment Measures Mean S.D. Attachment Measures Adolescent Secure Attachment Organization .26 .35 Adolescent Preoccupied Attachment Organization .07 .23 Adolescent Dismissing Attachment Organization .09 .38 Maternal Secure Attachment Organization .26 .37 Maternal Preoccupied Attachment Organization .09 .23 Maternal Dismissing Attachment Organization -.04 .35 Attachment and Autonomy in Adolescence 20 Attachment and Autonomy in Adolescence 21 Table 2 Means and Standard Deviations of Global Autonomy and Relatedness Scales Time 1 Time 2 Time 2 Time 1 Mean S.D. Mean S.D. t Autonomy Scales Adolescent Promoting Autonomy 1.92 .92 2.00 .80 .74 Mother Promoting Autonomy 2.71 .70 2.56 .62 -1.96* Adolescent Undermining Autonomy .89 .58 .93 .57 .58 Mother Undermining Autonomy .87 .46 1.18 .53 4.98*** Relatedness Scales Adolescent Promoting Relatedness 1.30 .52 1.32 .52 .19 Mother Promoting Relatedness 2.14 .54 1.89 .50 -4.42*** Adolescent Undermining Relatedness 1.20 .69 1.20 .63 3.71*** Mother Undermining Relatedness .90 .56 1.09 .58 2.92** Note. *** p < .001. ** p < .01. * p ≤ .05. + p < .10. N = 88. Attachment and Autonomy in Adolescence 22 Table 3 Means and Standard Deviations of Specific Autonomy and Relatedness Subscales Time 1 Time 2 Time 2-Time 1 Mean S.D. Mean S.D. t Autonomy Subscales: Adolescent Promoting Autonomy Adolescent Reasons 1.51 .83 1.51 .66 -.01 Adolescent Confidence 2.33 1.14 2.49 1.05 1.21 Mother Promoting Autonomy Mother Reasons 2.32 .84 2.11 .65 -2.20* Mother Confidence 3.09 .73 3.00 .73 -1.12 Adolescent Undermining Autonomy Adolescent Recanting 1.05 .77 .51 .64 -5.33*** Adolescent Blurring .82 .95 1.22 .98 2.99** Adolescent Pressuring .78 .87 1.06 .85 2.68** Mother Undermining Autonomy Mother Recanting .10 .26 .17 .40 1.52 Mother Blurring .98 .82 1.61 .93 5.56*** Mother Pressuring 1.53 .90 1.75 .89 2.20* Relatedness Subscales: Adolescent Promoting Relatedness Adolescent Queries 1.10 .85 .96 .69 -1.31 Adolescent Validating .94 .63 1.02 .77 .95 Adolescent Engagement 1.88 .80 1.96 .69 .93 Mother Promoting Relatedness Mother Queries 2.39 .73 1.80 .89 -6.04*** Mother Validating 1.52 .81 1.40 .77 -1.29 Mother Engagement 2.52 .68 2.47 .56 -.67 Adolescent Undermining Relatedness Adolescent Disrupting/Interrupting 1.73 .79 1.73 .74 -.02 Adolescent Hostile/Critical Statements .67 .79 .67 .77 -.03 Mother Undermining Relatedness Mother Disrupting/Interrupting 1.22 .78 1.39 .72 1.92+ Mother Hostile/Critical Statements .58 .62 .80 .74 2.47* Attachment and Autonomy in Adolescence 23 Note. *** p < .001. ** p < .01. * p < .05. + p < .10. N = 88. Attachment and Autonomy in Adolescence 24 Descriptive statistics. All measures were examined for possible gender, racial/ethnic minority status, and family income effects. Gender effects were not found for autonomy and relatedness behaviors, but significant effects emerged with attachment strategies (see Table 4). As hypothesized, female adolescents were more likely to exhibit a preoccupied attachment style than male adolescents (r = .29, p < .01). Also, consistent with prior findings, male adolescents tended to display dismissing attachment strategies more often than their female counterparts (r = .18, p <.10). Racial/ethnic minority status (a dummy variable coded as 0 = European-American; 1 = Member of Racial/Ethnic Minority Group) and family income both had significant effects on autonomy and relatedness behaviors as well as attachment style (see Table 4). Specifically, white, middle-class families holding secure attachment models and promoting autonomy and relatedness more often than their less privileged counterparts. Due to their effects on the examined measures, gender, racial/ethnic status, and family income were incorporated in all further analyses. Attachment and Autonomy in Adolescence 25 Table 4 Correlations of Autonomy and Relatedness Behaviors and Attachment Strategies with Gender, Racial/Ethnic Minority Status, and Family Income Gender Racial Minority Family Income r r r Attachment Organizations Adolescent Secure Attachment Organization .06 -.34*** .30** Adolescent Preoccupied Attachment Organization .29** -.09 -.17 Adolescent Dismissing Attachment Organization -.18+ .40*** -.31** Maternal Secure Attachment Organization -.04 -.30** .31** Maternal Preoccupied Attachment Organization -.17 -.03 -.09 Maternal Dismissing Attachment Organization .12 .41*** -.33*** Time 1: Autonomy Scales Adolescent Promoting Autonomy .04 -.31** .33*** Mother Promoting Autonomy -.08 -.05 .20+ Adolescent Undermining Autonomy .17 -.07 .02 Mother Undermining Autonomy -.17 .04 .05 Relatedness Scales Adolescent Promoting Relatedness -.08 -.19+ .24* Mother Promoting Relatedness -.03 -.26** .29** Adolescent Undermining Relatedness .02 -.09 -.05 Mother Undermining Relatedness -.14 -.05 .05 Time 2: Autonomy Scales Adolescent Exhibiting Autonomy .03 -.12 .20+ Mother Promoting Autonomy .05 .03 .08 Adolescent Undermining Autonomy .16 -.10 .02 Mother Undermining Autonomy -.11 .07 .03 Relatedness Scales Adolescent Promoting Relatedness -.09 -.04 .14 Mother Promoting Relatedness -.05 -.06 .20+ Adolescent Undermining Relatedness .08 -.16 .03 Mother Undermining Relatedness .04 .14 -.13 Attachment and Autonomy in Adolescence 26 Note. *** p≤ .001. ** p < .01.* p < .05.+ p < .10. Attachment and Autonomy in Adolescence 27 Intercorrelations among attachment measures. Correlations between adolescent and maternal attachment organizations are presented in Table 5. The scales measuring security of attachment representation were highly negatively correlated with the dismissing scales for both adolescents (r = -.93, p < .001) and mothers (r = -.85, p < .001) The scales for security were also moderately negatively correlated with the preoccupation scales for adolescents (r = -.37, p <.001) and mothers (r = -.51, p < .001). Preoccupation scales were also positively correlated with dismissing scales (r = .19, p <.10 for adolescents; r = .34, p <.001 for mothers). Because of the redundancy indicated by the high correlations among the security and dismissing scales, the maternal and adolescent dismissing scales were not used in further analyses. Adolescent attachment organization was not related to maternal attachment organization beyond a trend level: adolescent secure attachment organization correlated negatively with maternal preoccupied attachment organization (r = -.19, p < .10) and adolescent dismissing attachment organizations correlated positively with maternal preoccupied attachment organization (r = .19, p < .10). Table 5 Intercorrelations among Attachment Organizations Attachment Organizations 1. 2. 3. 4. 5. 6. 1. Adolescent Secure -2. Adolescent Preoccupied -.37*** -3. Adolescent Dismissing -.93*** .19+ -4. Maternal Secure .15 .01 -.15 -5. Maternal Preoccupied -.19+ .13 .19+ -.51*** -6. Maternal Dismissing -.13 -.03 .11 -.85*** .34*** -Note. *** p< .001. *** p < .01.* p < .05.+ p < .10. Attachment and Autonomy in Adolescence 28 Attachment and Autonomy in Adolescence 29 Intercorrelations among autonomy and relatedness measures. Correlations between each global autonomy and relatedness scale are presented in Table 6. Results show that interactive behaviors promoting autonomy and relatedness were positively related for adolescents at ages 16 and 18 (r = .25, p < .05 for Time 1; r = .49, p ≤ .001 for Time 2) and for mothers when adolescents were 16 only (r = .28, p ≤ .01 for Time 1). The data also reveal that undermining autonomy was positively correlated with undermining relatedness for both adolescents (r = .45, p ≤ .001 for Time 1; r = .60, p ≤ .001 for Time 2) and mothers (r = .43, p ≤ .001 for Time 1; r = .58, p ≤ .001 for Time 2). Results also indicate that adolescent global scales for autonomy and relatedness were related to maternal scales (see Table 6). Intercorrelations among each specific autonomy subscale are presented in Table 7a and those among specific relatedness subscales are presented in Table 7b. These results indicate similar relations between adolescent and maternal behaviors emerge when global autonomy and relatedness scales are broken down into their specific subscales (see Table 7a & 7b). Attachment and Autonomy in Adolescence 30 Attachment and Autonomy in Adolescence 31 Attachment and Autonomy in Adolescence 32 Attachment and Autonomy in Adolescence 33 Primary Analyses Simple correlational analyses of attachment organizations and global autonomy and relatedness scales. Initial analyses looked at simple correlations between maternal and adolescent attachment security and preoccupation and global measures of autonomy and relatedness as well as specific individual autonomy and relatedness behaviors. These simple correlations are presented in Table 8. They indicate numerous relationships between attachment models and behaviors displaying and promoting autonomy and relatedness and a few links with undermining behaviors. These relationships will be explored further below. Table 8 Simple Correlations of Global Autonomy and Relatedness Scales with Attachment States of Mind Security Preoccupation Adolescent Maternal Adolescent Maternal r r r r Time 1: Autonomy Scales Adolescent Promoting Autonomy .09 .31** .11 -.19+ Mother Promoting Autonomy -.06 .13 -.00 -.01 Adolescent Undermining Autonomy -.06 .02 .04 -.13 Mother Undermining Autonomy -.07 .13 -.05 .02 Relatedness Scales Adolescent Promoting Relatedness .35*** .29 -.19+ -.35*** Mother Promoting Relatedness .19+ .29** .14 -.07 Adolescent Undermining Relatedness -.18+ .04 .22* -.02 Mother Undermining Relatedness .06 .05 -.02 -.02 Time 2: Autonomy Scales Adolescent Promoting Autonomy .02 .07 .06 .04 Mother Promoting Autonomy .06 .17 -.08 .00 Adolescent Undermining Autonomy -.14 -.10 .08 .14 Mother Undermining Autonomy -.08 -.15 -.14 .06 Relatedness Scales Adolescent Promoting Relatedness .09 .26* .06 -.05 Mother Promoting Relatedness .15 .19+ .02 -.10 Adolescent Undermining Relatedness -.12 -.10 .16 .21* Attachment and Autonomy in Adolescence 34 Mother Undermining Relatedness -.05 -.11 .02 .05 Note. *** p < .001. ** p < .01. * p < .05. + p < .10. Attachment and Autonomy in Adolescence 35 Correlation analyses of attachment organizations and autonomy and relatedness. The first set of analyses describe the most typical behaviors exhibited by adolescents and their mothers when one member of the dyad displayed a more secure or more preoccupied attachment style. Thus these data answer the question: how did security or preoccupation manifest itself in family interactions around autonomy and relatedness? Primary analyses for this question used partial correlations between adolescent and maternal attachment organizations and global measures of autonomy and relatedness, after accounting for the effects of the three demographic variables previously identified (a dummy variable for gender, a dummy variable for membership in a racial/ethnic minority group, and family income). These data are presented in Tables 9, and, as expected, indicate that attachment security and preoccupation are related to a wide range of family members’ behaviors in handling issues of autonomy and relatedness. Attachment and Autonomy in Adolescence 36 Table 9 Correlations of Autonomy and Relatedness Behaviors with Attachment States of Mind (After Partialing Gender, Race/Ethnic Status, and Family Income) Adolescent Maternal Adolescent Maternal Security Security Preoccupation Preoccupation Partial r Partial r Partial r Partial r Time 1: Autonomy Scales Adolescent Promoting Autonomy -.06 .21+ .13 -.17 Mother Promoting Autonomy -.11 .08 .06 .00 Adolescent Undermining Autonomy -.11 .00 -.01 -.11 Mother Undermining Autonomy -.06 .14 .02 .01 Relatedness Scales Adolescent Promoting Relatedness .30** .23* -.17 -.36*** Mother Promoting Relatedness .08 .20+ .19+ -.06 Adolescent Undermining Relatedness -.22* .03 .20+ -.04 Mother Undermining Relatedness .05 .04 .02 -.05 Time 2: Autonomy Scales Adolescent Promoting Autonomy .08 .01 .08 .07 Mother Promoting Autonomy .05 .17 -.07 .03 Adolescent Undermining Autonomy -.20+ -.12 .02 .17 Mother Undermining Autonomy -.06 -.15 -.07 .06 Relatedness Scales Adolescent Promoting Relatedness .06 .24* .11 -.05 Mother Promoting Relatedness .11 .15 .07 -.08 Adolescent Undermining Relatedness -.19+ -.15 .12 .22* Mother Undermining Relatedness .01 -.06 .01 .05 Note. *** p < .001. ** p ≤ .01.* p < .05.+ p < .10. Attachment and Autonomy in Adolescence 37 As hypothesized, maternal secure attachment organization was positively related to adolescent promotion of relatedness at ages sixteen and eighteen (r = .23, p < .05 for Time 1; r = .24, p < .01 for Time 2). Results also showed that adolescent secure attachment representations related positively to adolescents exhibiting relatedness at age 16 (r = .30, p < .01). In addition, adolescent secure attachment organizations were negatively related to adolescents’ undermining autonomy at ages 16 and 18 ( r = -.22, p < .05 for Time 1; r = -.19, p < .10 for Time 2). Maternal preoccupation with attachment was found to be negatively correlated with adolescents’ exhibition of relatedness at age 16 (r = -.36, p < .001). At age 18, adolescents with more preoccupied mothers attempted to undermine relatedness (r = .22, p < .05). Adolescent preoccupation with attachment showed a trend towards positively relating to mothers’ promoting relatedness (r = .19, p < .10) and adolescents’ undermining of autonomy (r = .20, p < .10) when adolescents were 16 years old. No significant findings, however, were observed between adolescents’ preoccupied attachment models and family behaviors two years later (see Table 9). In order to further explore possible ways in which attachment models manifest themselves in family interactions, the specific subscales were investigated. However, the relations between the subscales and attachment models must be interpreted with caution since the sheer number of subscales being used in analyses increase the probability that chance relationships will surface. Yet, analyses still revealed that attachment models were related to specific interaction behaviors in similar patterns to those involving global measures of autonomy and relatedness. These findings are presented in Appendix B. Attachment and Autonomy in Adolescence 38 Hierarchical regression analyses of attachment organizations and measures of autonomy and relatedness. The second set of primary analyses addressed how different aspects of mothers’ and adolescents’ attachment models combine to explain autonomy and relatedness behaviors at a given point in time. Hierarchical regression analyses examined the main effects of gender, racial/ethnic minority status, family income, adolescent security, maternal security, adolescent preoccupation, followed by maternal preoccupation in predicting global as well as specific measures of autonomy and relatedness. Variables were entered in the above order. Relationships between attachment and all scales and subscales were tested. Significant findings are presented in Tables 10-13 and are discussed in further detail below. In examining the degree to which adolescents exhibited autonomy at age 16, analyses revealed that after accounting for the effects of related demographic covariates, only maternal secure attachment organization positively predicted for adolescent autonomy at a trend level or above (β = .21, p < .10) within an overall significant model (see Table 10). Table 10 Hierarchical Regressions Predicting Adolescent Autonomy at Time 1 from Adolescent and Maternal Attachment Organizations After Accounting for Related Covariates Adolescent Autonomy at Time 1 I. β Total R ∆R Gender (1=M; 2=F) .08 Race (0=White; 1=African American/Other) -.18 Family Income .26* .145** .145 II. Adolescent Security III. -.06 .147** .002 Maternal Security .21+ .185** .038+ IV. Adolescent Preoccupation V. .12 .195** .010 Maternal Preoccupation .11 .203** .008 Attachment and Autonomy in Adolescence 39 Note. *** p < .001. ** p < .01. * p < .05. + p < .10. β weights are those taken from entry of variables into models. N = 88. Attachment and Autonomy in Adolescence 40 Further analyses indicated that adolescent positive relatedness at age 16 can be predicted from a combination of attachment representations (see Table 11). Specifically, adolescent security (β = .31, p < .01), maternal security (β = .23, p < .05), and maternal preoccupation (β = -.28, p <.05) each were significant predictors of adolescent relatedness after accounting for the effects of demographic variables, accounting for 17.6% of the variance, combined. Table 11 Hierarchical Regressions Predicting Adolescent Relatedness at Time 1 from Maternal and Adolescent Attachment Organizations After Accounting for Related Covariates Adolescent Relatedness at Time 1 I. β Total R ∆R Gender (1=M; 2=F) -.05 Race (0=White; 1=African American/Other) -.09 Family Income .18 .065 .065 II. Adolescent Security III. .31** .147** .082** Maternal Security .23* .192** .045* IV. Adolescent Preoccupation V. -.06 .194** .002 Maternal Preoccupation -.28* .243** .049* Note. *** p < .001. ** p < .01. * p < .05. + p < .10. β weights are those taken from entry of variables into models. N = 88. Attachment and Autonomy in Adolescence 41 When making predictions from attachment representations as assessed when adolescents were 16 to interactive behaviors at age 18, only maternal secure attachment organization remained as a significant predictor of adolescent relatedness (β = .25, p < .05) (see Table 12). Table 12 Hierarchical Regressions Predicting Adolescent Relatedness at Time 2 from Adolescent and Maternal Attachment Organizations After Accounting for Related Covariates Adolescent Relatedness at Time 2 I. β Total R ∆R Gender (1=M; 2=F) -.07 Race (0=White; 1=African American/Other) .03 Family Income .14 .025 .025 II. Adolescent Security III. .07 .029 .004 Maternal Security .25* .084 .059* IV. Adolescent Preoccupation V. .18 .105 .021 Maternal Preoccupation .12 .114 .009 Note. *** p < .001. ** p < .01. * p < .05. + p < .10. β weights are those taken from entry of variables into models. N = 88. Attachment and Autonomy in Adolescence 42 Analyses also showed that maternal positive relatedness when adolescents were 16 could be predicted by adolescent preoccupation, with a trend toward maternal security also contributing (see Table 13). Specifically, maternal security (β = .20, p < .10) and adolescent preoccupation (β = .28, p < .05) account for a combined 9.0% of the variance after the effects of related covariables were considered. Table 13 Hierarchical Regressions Predicting Maternal Relatedness at Time 1 from Maternal and Adolescent Attachment Organizations After Accounting for Related Covariates Maternal Relatedness at Time 1 I. β Total R ∆R Gender (1=M; 2=F) .01 Race (0=White; 1=African American/Other) -.15 Family Income .22+ .103* .103 II. Adolescent Security III. .09 .109* .006 Maternal Security .20+ .143* .036+ IV. Adolescent Preoccupation V. .28* .197** .054* Maternal Preoccupation .05 .199** .002 Note. *** p < .001. ** p ≤ 01. * p < .05. + p < .10. β weights are those taken from entry of variables into models. N = 88. All specific subscales were once again investigated to further explore the relationships between attachment and interaction behaviors. Significant findings, presented in Appendix C, largely replicate the Attachment and Autonomy in Adolescence 43 patterns found for the global autonomy and relatedness scales. Analyses still revealed that interaction behaviors often appeared to be jointly determined by each individual’s attachment organization. Hierarchical regression analyses of attachment models and changes over time of global measures of autonomy and relatedness. The relative changes in autonomy and relatedness behaviors over the two-year period were determined based on the combined effects of adolescents’ and mothers’ states of mind regarding attachment. More specifically, in order to identify which adolescents gained the most and least autonomy and relatedness over the two years following the initial assessment, a hierarchical regression strategy, similar to that described above, was used in which future levels of autonomy and relatedness (as indicated at age 18) were predicted after first accounting for current levels (those at age 16) and other relevant factors. Predicting a future level of a variable while accounting for predictions from initial levels (e.g., stability) yields a series of path models which reflect one type of change over time: increases or decreases in the observed variable relative to predictions based upon initial levels (Cohen & Cohen, 1975). These analyses revealed no significant predictions of change to global scales of behaviors around autonomy and relatedness, although there were two trends. Specifically, maternal secure attachment organization displayed a trend toward predicting adolescent relatedness at age 18 (β = .20, p < .10) even after accounting for the effects of related factors and adolescent relatedness at age 16, and maternal preoccupation displays a trend toward predicting adolescent negative relatedness at age 18 (β = .20, p < .10) after accounting for related demographic factors and the level of negative relatedness exhibited by adolescents at age 16. Attachment and Autonomy in Adolescence 44 Similar predictions were observed for various specific family behaviors, although, once again interpretations of these findings must be made with caution. For instance, analyses revealed that the degree to which adolescents pressured their mothers at age 18 was predicted by security of adolescent attachment models (β = -.27, p ≤ .01) combined with security of maternal attachment models (β = .18, p < .10). Again, these predictions are those that existed above and beyond what was predicted from related factors and the degree to which adolescents pressured two years earlier (see Table 14). Table 14 Hierarchical Regressions Predicting Adolescent Pressuring at Time 2 from Adolescent and Maternal Attachment Organizations After Accounting for Adolescent Pressuring at Time 1 and Related Covariates Adolescent Pressuring at Time 2 β R Total ∆R I. Gender (1=M; 2=F) .11 Race (0=White; 1=African American/Other) -.14 Family Income .03 .033 .033 II. Adolescent Pressuring at Time 1 .36*** .162** .129*** III. Adolescent Security IV. Maternal Security -.27** -.18+ .222*** .251*** .060*** .049+ V. Adolescent Preoccupation VI. Maternal Preoccupation -.08 .16 .255*** .272*** .004 .017 Note. *** p ≤ .001. ** p ≤ .01. * p < .05. + p < .10. β weights are those taken from entry of variables into models. N = 88. Attachment and Autonomy in Adolescence 45 Results also indicated that mothers’ secure attachment models significantly predicted the level and extent of their reasoning when their adolescents were 18 beyond what was predicted by related covariables and their level of reasoning when adolescents were 16 (β = .27, p < .05) (see Table 15). Table 15 Hierarchical Regressions Predicting Mothers’ Reasoning at Time 2 from Adolescent and Maternal Attachment Organizations After Accounting for Mothers’ Reasoning at Time 1 and Related Covariates Mother Reasons at Time 2 β R Total ∆R I. Gender (1=M; 2=F) .03 Race (0=White; 1=African American/Other) .01 Family Income .09 .007 .007 II. Mother Reasons at Time 1 .30** .093+ .086** III. Adolescent Security IV. Maternal Security .12 .27* .105+ .166* .012 .061* V. Adolescent Preoccupation VI. Maternal Preoccupation -.08 .15 .171* .185* .005 .014 Note. *** p < .001. ** p < .01. * p < .05. + p < .10. β weights are those taken from entry of variables into models. N = 88. In addition, maternal preoccupied attachment representation significantly predicted the quality and amount of questions posed by adolescents at age 18 (β = .28, p ≤ .05) after accounting for the effects of related covariables and level of questioning at age 16 (see Table 16). Attachment and Autonomy in Adolescence 46 Table 16 Hierarchical Regressions Predicting Adolescent Query Behavior at Time 2 from Adolescent and Maternal Attachment Organizations After Accounting for Adolescent Query Behavior at Time 1 and Related Covariates Adolescent Queries at Time 2 β R Total ∆R I. Gender (1=M; 2=F) -.11 Race (0=White; 1=African American/Other) .00 Family Income -.07 .016 .016 II. Adolescent Queries at Time 1 .21* .059 .043* III. Adolescent Security IV. Maternal Security -.00 -.03 .059 .060 .000 .001 V. Adolescent Preoccupation VI. Maternal Preoccupation -.04 .28* .061 .111 .001 .050* Note. *** p < .001. ** p < .01. * p ≤ .05. + p < .10. β weights are those taken from entry of variables into models. N = 88. Further analyses revealed that the degree to which older adolescents validated their mothers thoughts and opinions was predicted from the combined effects of maternal and adolescent attachment states of mind (see Table 17). Specifically, maternal secure attachment organization (β = .32, p ≤ .01) combined with adolescent preoccupied attachment organization (β = .26, p < .05) to predict the degree to which adolescents validated their mothers at age 18 even after accounting for the predictive effects of related demographic variables and the amount of validating exhibited by adolescents two years earlier. Attachment and Autonomy in Adolescence 47 Table 17 Hierarchical Regressions Predicting Adolescent Validating Behavior at Time 2 from Adolescent and Maternal Attachment Organizations After Accounting for Adolescent Validating Behavior at Time 1 and Related Covariates Adolescent Validating at Time 2 β R Total ∆R I. Gender (1=M; 2=F) -.00 Race (0=White; 1=African American/Other) .03 Family Income .18 .028 .028 II. Adolescent Validating at Time 1 .25* .087+ .059* III. Adolescent Security IV. Maternal Security -.01 .32** .087 .177** .000 .090** V. Adolescent Preoccupation VI. Maternal Preoccupation .26* .01 .222** .222** .045* .000 Note. *** p < .001. ** p ≤ .01. * p < .05. + p ≤ .10. β weights are those taken from entry of variables into models. N = 88. Attachment and Autonomy in Adolescence 48 Results also showed that mothers with more secure models of attachment were likely to increase the degree to which they were engaged in the discussions (β = .23, p < .05) more than other mothers even after accounting for the effects of their level of engagement when their children were 16 years old (see Table 18). Table 18 Hierarchical Regressions Predicting Maternal Engagement at Time 2 from Adolescent and Maternal Attachment Organizations After Accounting for Maternal Engagement at Time 1 and Related Covariates Maternal Engagement at Time 2 β R Total ∆R I. Gender (1=M; 2=F) .06 Race (0=White; 1=African American/Other) -.08 Family Income .08 .022 .022 II. Maternal Engagement at Time 1 .34** .128* .106** III. Adolescent Security IV. Maternal Security .09 .23* .134* .178** .006 .044* V. Adolescent Preoccupation VI. Maternal Preoccupation .15 .02 .193** .194* .015 .001 Note. *** p < .001. ** p ≤.01. * p < .05. + p < .10. β weights are those taken from entry of variables into models. N = 88. Attachment and Autonomy in Adolescence 49 Final analyses revealed that maternal attachment states of mind also predicted changes in adolescents’ levels of critical/hostile behavior from age 16 to age 18 (see Table 19). Specifically, those adolescents who became most hostile over time were more likely to have mothers who adhered to preoccupied attachment models (β = .24, p < .05), while those adolescents whose level of hostility decreased the most over time had mothers who held more secure attachment representations (β = -.27, p < .01). Table 19 Hierarchical Regressions Predicting Adolescent Critical or Hostile Behavior at Time 2 from Adolescent and Maternal Attachment Organizations After Accounting for Adolescent Critical or Hostile Behavior at Time 1 and Related Covariates Adolescent Critical/Hostile at Time 2 β R Total ∆R I. Gender (1=M; 2=F) .10 Race (0=White; 1=African American/Other) -.28* Family Income -.12 .071+ .071+ II. Adolescent Relatedness at Time 1 .42*** .236*** .165*** III. Adolescent Security IV. Maternal Security -.13 -.27** .250*** .316*** .014 .066** V. Adolescent Preoccupation VI. Maternal Preoccupation -.09 .24* .322*** .358*** .006 .036* Note. *** p < .001. ** p < .01. * p < .05. + p < .10. β weights are those taken from entry of variables into models. N = 88. Attachment and Autonomy in Adolescence 50 Attachment and Autonomy in Adolescence 51 Discussion The current study provided further support for the idea that attachment models are linked to behaviors involving autonomy and relatedness in adolescence. As hypothesized, both adolescents and mothers with more secure attachment representations tended to promote autonomy while maintaining relatedness, whereas preoccupied family members tended to undermine autonomy while both promoting and undermining relatedness. Expanding on previous research, results also showed that the degree to which families successfully maintained relatedness while promoting autonomy can best be predicted by both family members’ attachment states of mind rather than by any given individual attachment model. In addition, the present study further revealed that the attachment representations held by adolescents and their mothers predicted relative changes in how families handle the adolescent struggle for autonomy over time. Each of these findings will be discussed in greater detail below. Predictions from Attachment Security As hypothesized, the attachment representations of adolescents and their mothers were significantly related to how family members interacted during family discussions. Each individual’s model of attachment corresponded to specific adolescent and maternal behaviors that either promoted or undermined autonomy and/or relatedness. These findings are summarized in Table 20 below. Attachment and Autonomy in Adolescence 52