Fathers’ Contributions to Attachment in Adolescence and Adulthood: The Moderating Role of Race, Gender, Income, and Residential Status

Abstract

Fathers play a critical yet underappreciated role in adolescent social development. To examine contributions of fathers’ parenting to attachment security in adolescence and adulthood, the present longitudinal study followed 184 adolescents (99 females, 85 males; 58% White, 29% Black, 13% other racial identities) from ages 13–24. At age 13, adolescents reported on their fathers’ parenting behavior and were observed in a father–teen conflict task; at ages 14 and 24, they completed the Adult Attachment Interview. Adolescents who lived with their father showed significantly higher attachment security at age 14 (Cohen’s d = .72, a medium-to-large effect size), compared to those with non-residential fathers. Fathers’ observed positive relatedness with teens and adolescents’ perceptions of fathers’ support for their psychological autonomy vs. control at age 13 predicted attachment security at age 14. Fathers’ physical aggression predicted lower attachment security in adolescence, whereas fathers’ verbal aggression uniquely predicted lower security in adulthood, illuminating developmental shifts. Pathways to security were moderated by father residential status, adolescent gender, and racial identity. Findings underscore the importance of fathers’ presence, autonomy support, and non-aggression in predicting short- and long-term outcomes for adolescents’ state of mind in close relationships.

Fathers play a critical yet underappreciated role in child and adolescent development (Cabrera et al., 2018). In adolescence, fathers’ presence in their teens’ lives is associated with reduced risk for behavior problems and delinquency, and with positive social and relational functioning in adulthood (see East et al., 2006; and Sarkadi et al., 2008). Moreover, some research suggests that high-quality father–adolescent relationships may be even more important for teens’ social competence (Rice et al., 1997) and reduced depressive symptoms (Pan et al., 2006) than the mother-child relationship.

From an attachment perspective (Bowlby, 1969/1982), one potential mechanism linking experiences of paternal care to positive developmental outcomes is a secure internal working model of attachment—that is, a view of the self as worthy of love and a sense of self-efficacy in the world, of others as generally trustworthy, and of relationships as valued assets in one’s life (Main et al., 1985). Over the past two decades, researchers have made substantial advances in understanding the role of paternal care in shaping attachment behavior in early childhood (e.g., Bureau et al, 2017). Here we extend this work into adolescence, by examining multiple dimensions of paternal care—including fathers’ physical presence (i.e., residential status), autonomy support, and threat behavior—as potential “active ingredients” in the development of internal working models from adolescence through early adulthood.

The role of fathers in child attachment

A core tenet of attachment theory is that children form attachment bonds to one or more caregivers who provide support for autonomous exploration and safety in times of threat— that is, a secure base (Bowlby, 1969/1982). At the species level, cooperative caregiving is the norm (Hrdy, 2009), such that most children are embedded in a network of multiple attachment figures (Howes & Spieker, 2016); yet within and across cultures, there is substantial variability in the extent to which fathers are involved in childrearing (Hewlett, 2000; Lamb, 2013). When fathers are involved in caregiving, Bowlby observed that children can and do form attachment bonds with fathers that contribute to secure internal working models of the self and others: “A young child’s experience of an encouraging, supportive, and cooperative mother, and a little later father, gives him a sense of worth, a belief in the helpfulness of others, and a favorable model on which to build future relationships” (Bowlby, 1982, p. 378).

A growing body of research has examined paternal behaviors in infancy and early childhood that contribute to secure child attachment (e.g., Bureau et al., 2017). For example, early father involvement and sensitivity show unique contributions to children’s secure attachment to fathers at age 3 (Brown et al., 2007). Longitudinal evidence shows that fathers’ sensitive and challenging interactive play in the toddler years is linked to secure internal working models of attachment in middle childhood and adolescence (Grossmann et al., 2002). In turn, children’s secure attachment to fathers has been shown to have unique and interactive effects with mother-child attachment in predicting a host of developmental outcomes in childhood (see Dagan & Sagi-Schwartz, 2021; Dagan et al., 2021; Deneault et al., 2021).

Yet we know comparatively little about the paternal behaviors in adolescence that predict secure attachment during this “second sensitive period” of social development (Blakemore & Mills, 2014). Teens’ changing attachment needs and rapidly developing social competencies call for major adaptations in caregiving behavior, including a changing role for fathers as children age (Lamb & Lewis, 2014). Whereas sensitive and challenging play is vital in early childhood (Grossmann et al., 2002), new forms of paternal care may be needed for fathers to provide an effective secure base in adolescence.

Attachment in adolescence: Possible shifts in fathers’ role as a secure base

Developmentally, at least three important shifts in the attachment system take place during adolescence, with implications for the role of fathers. First, as teens grow in their social-cognitive competencies and spend more time with peers, their dependence on parents decreases, while the desire for exploration and independence increases (Allen & Tan, 2016). This normative shift toward autonomy-seeking typically involves a renegotiation of attachment relationships with parents, which can increase conflict. For example, a teen might renegotiate household rules about how late and with whom they can stay out on a Saturday night, igniting an argument with their father, whose goal of safety for his child may conflict with the teen’s goals of exploration and enjoying a good time with friends.

Although changes in autonomy-seeking and conflict are normative in adolescence, meaningful differences in how these changes are negotiated with parents—including fathers—have important implications for teens’ representations of attachment relationships. Theory suggests that fathers play a unique role in supporting children’s autonomy and encouraging exploration and risk-taking (e.g., Paquette’s [2004] “father-child activation relationship”; see also Richard Bowlby’s “explore and excite” model of fathers as attachment figures [Newland & Coyl, 2010]). For example, scholars have suggested that whereas mothers serve as a secure base for teens to process emotions and interpersonal experiences, fathers often take on the role of helping teens explore future-oriented goals and experiences in the wider world (Shulman & Seiffge-Krenke, 1997). Thus, it is reasonable to expect that fathers’ autonomy-supportive paternal care would be particularly relevant to attachment security. One previous study found that father–teen conflict interactions that balance autonomy and relatedness (observed at the dyadic level) were concurrently associated with adolescent attachment security (Allen et al., 2007); related work in another sample found that autonomy and relatedness with both parents at age 14 predicted security at age 25 (Allen & Hauser, 1996). Conversely, paternal behaviors that undermine autonomy, such as psychological control, may be especially detrimental to adolescent attachment security. Indeed, prior work suggests that fathers’ use of psychological control is concurrently associated with lower self-reported attachment security in a sample of adolescents and young adults (Leondari & Kiosseoglou, 2002).

Second, during adolescence, internal working models of attachment to a specific caregiver are thought to generalize to broader representations of close relationships, which can be measured using the Adult Attachment Interview (AAI; George et al., 1996). The AAI asks about experiences with both fathers and mothers but is typically scored in terms of an individual’s overall state of mind regarding attachment, rather than attachment to an individual caregiver. Markers of a secure state of mind include narrative coherence, emotion regulation, emotional balance, autonomy, and valuing close relationships (Hesse, 2016). Thus, it is possible to test whether paternal care matters not just for attachment to fathers specifically, but also for broader representations of the self and others generally—a key mechanism linking caregiving experiences to long-term outcomes with peers, romantic partners, and one’s own future children (Borelli et al., 2019; van IJzendoorn, 1995). Interestingly, data comparing the AAI to questionnaire measures of adult attachment reveal that father-related AAI variables predict women’s self-reported romantic attachment style better than mother-related AAI variables (Shaver et al., 2020).

Third, researchers have characterized adolescent attachment on the AAI as reflecting teens’ emotion regulation (Allen & Miga, 2010), as assessed via the capacity to remain regulated while discussing difficult relationship topics (such as loss). As experiences of emotion intensify during puberty (Bailen et al., 2019), parents’ role as secure bases shifts toward balancing teens’ autonomy needs with ongoing availability to co-regulate emotion—and conversely, to avoid behaviors that threaten or dysregulate the teen. Research in early childhood has identified several ways in which paternal care contributes to children’s emotion regulation capacities, including fathers’ presence, operationalized as co-habitation (Sarkadi et al., 2008) and consistency of father presence (Bocknek et al., 2014). In adolescence, father acceptance has been found to predict adolescent girls’ emotion regulation development one year later, above and beyond teens’ age, baseline regulatory skills, depressive risk, and maternal acceptance (Vergara-Lopez et al., 2022).

Conversely, fathers’ aggression during conflict, unsupportive responses to children’s expression of emotion, and modeling of dysregulation and hostility have been linked to emotion regulation problems in children and adolescents (for reviews see Islamiah et al., 2023; Kiel & Kalomiris, 2015). One study found that fathers’ harsh parenting at age 13–14 was concurrently associated with insecure attachment on the AAI (Allen et al., 2007). Other work has shown that paternal verbal aggression during conflict may play a role in predicting teens’ behavior in future social relationships with peers (Bailey et al., 2024) and romantic partners (Allen et al., 2024). Thus, adolescents’ perception of threat within the father–teen relationship is likely to undermine their self-regulation and sense of safety.

Critically, however, very few longitudinal studies have examined the paternal behaviors that predict secure attachment representations in adolescence, and whether these links are similar or different in early adulthood. On the one hand, the impact of fathers’ parenting behaviors on attachment in adolescence may be minimal, given the substantial role of peer relationships (Allen & Miga, 2010) and the tendency for mothers to serve as primary attachment figures during this developmental period (Rosenthal & Kobak, 2010). On the other hand, father behavior may have equal or greater weight in predicting attachment in early adulthood (e.g., Matsuoka et al., 2006), given that fathers’ involvement in childrearing tends to increase with age (Bailey, 1994), and that certain paternal behaviors show unique, additive, and in some cases stronger effects than maternal behavior on attachment-related outcomes, such as adolescent conflict with friends (Lieberman et al., 1999), depressive symptoms (Pan et al., 2006), and social competence (Forehand & Nousiainen, 1993; Rice et al., 1997).

Fatherhood in context: Examining diversity of developmental pathways

Presuming that fathers do matter for adolescent attachment, key questions remain regarding potential sources of variability in when and for whom specific paternal behaviors matter. In particular, it is not yet known whether father factors show developmental continuity or change in predicting attachment in adolescence vs. early adulthood, or whether pathways vary as a function of family and adolescent characteristics. Beyond the potential main effects of fathers’ presence, the contributions of specific paternal behaviors may be moderated by whether fathers live with teens in the home. Previous research has shown that associations between father–adolescent conflict and teens’ internalizing and externalizing problems are moderated by fathers’ residential status (Little et al., 2019), and that father involvement is more strongly associated with adolescent adjustment when fathers lived with the teen (Carlson, 2006). Related evidence in younger children shows that fathers’ negative caregiving behaviors are associated with lower child attachment security only when fathers are more involved in children’s daily lives (Brown et al., 2007).

In addition to residential status, the context in which father–adolescent interactions occur may also be shaped by families’ socioeconomic status, teens’ gender (Matsuoka et al., 2006; Zou et al., 2020), and racial-ethnic identity (Jones et al., 2024; Tyrell & Masten, 2022). For example, researchers have suggested that parents of Black youth may adopt a unique repertoire of caregiving behaviors to provide a secure base for children and adolescents to cope with racism-related stressors (Dunbar et al., 2017; Dunbar et al., 2022), yet little research has examined secure base provision among fathers of Black adolescents (see Tyrell & Masten, 2022). Thus, it is crucial to better characterize potential sources of diversity in fathers’ contributions to secure attachment (Stern et al., 2022).

The present study

The aim of the present study is to examine father contributions to secure representations of attachment in adolescence and early adulthood, as assessed on the AAI at age 14 and 24. We leverage a multimethod longitudinal dataset integrating observations of father–teen interactions, adolescent reports of father behavior, and semi-structured interviews spanning over a decade. Drawing on attachment theory, we examined a set of predictors in two domains: First, we examined paternal secure base support as a set of promotive factors predicting enhanced security, including (a) fathers’ presence (operationalized as living in the same household with the teen) and (b) support for autonomy and relatedness (operationalized as adolescents’ perceptions of fathers’ acceptance, psychological autonomy vs. control, and dyadic observations of fathers’ autonomy support and relatedness during conflict). Second, we examined paternal threat (operationalized as paternal physical and verbal aggression during conflict, as reported by the teen) as a risk factor predicting insecurity. Specifically, we test the following hypotheses:

1. Father residential status will predict adolescent attachment, such that living with one’s biological or adoptive father will predict greater security in mid-adolescence.

2. Replicating and extending previous findings (Allen & Hauser, 1996), fathers’ support for adolescents’ autonomy and relatedness—as observed in dyadic interactions and in teen reports of fathers’ acceptance and psychological autonomy (vs. control)—will predict attachment security in mid-adolescence and adulthood.

3. Threats to adolescents’ sense of safety with their fathers—specifically fathers’ use of physical and verbal aggression toward teens in early adolescence—will undermine attachment security in mid-adolescence and adulthood.

4. Importantly, we also predict that fathers will have unique contributions to attachment security in adolescence and adulthood, over and above the well-established contributions of mothers.

Note that concurrent associations of adolescent attachment with father autonomy/ relatedness and harsh parenting have been reported elsewhere (Allen et al., 2007). In the present investigation, we disaggregate specific dimensions of each paternal behavior (i.e., unique contributions of autonomy vs. relatedness; verbal vs. psychological aggression) to provide a more fine-grained understanding of their links to attachment and extend these findings to predict attachment 10 years later, in early adulthood. Moreover, in a novel extension of previous work, we conduct exploratory analyses to examine (a) whether father and mother predictors interact, to identify potential additive and compensatory effects; and (b) whether father contributions to attachment will be moderated by adolescent gender, race, father residential status, and family income. Given the characteristics of the present sample, as well as evidence for unique developmental processes in Black families to cope with racism (Dunbar et al., 2017; Stern et al., 2023), we focused our examination of racial identity on Black adolescents. Broadly, we predict that the effects of fathers’ behavior will be stronger for fathers who live with the teen; however, we have no a priori hypotheses regarding the other potential moderators.

Method

Participants

This report draws from an ongoing longitudinal investigation of adolescent social development in familial and peer contexts, beginning in 1998. Adolescents were originally recruited from the seventh and eighth grades of a public middle school drawing from suburban and urban populations in the mid-Atlantic United States. A final sample of 184 adolescents (85 males, 99 females) and their parents was first assessed at age 13 and reassessed annually. The sample was demographically diverse and representative of the community from which it was drawn: 107 adolescents (58%) identified as white, 53 (29%) as African American, 15 (8%) as of mixed race or ethnicity, and 9 (5%) as being from other identity groups. The present study utilizes data collected when teens were 13 (M = 13.35, SD = 0.64), 14 (M = 14.27, SD = 0.77), and 24 (M = 24.65, SD = 0.96).

At the initial assessment, 96 adolescents (52%) were living with their biological or adoptive fathers. Of the 110 fathers who attended the initial assessment (M_age = 44.76, SD = 6.35), most were employed (90%) and married to the teen’s mother (79%). Among the 88 non-residential fathers in the sample, 12 were married to the teen’s mother at the time of the initial assessment, 1 was single, and the remainder were separated or divorced. Adolescents’ parents reported a median family income in the $40,000–$59,999 range at the initial assessment.

Procedure

For all data collection, adolescents provided informed assent, and their parents provided informed consent before each interview session. Once participants reached age 18, they provided informed consent. In the initial introduction and throughout each session, confidentiality was explained to all family members, and adolescents were told that their parents would not be informed of any of the answers they provided. A Confidentiality Certificate, issued by the U.S. Department of Health and Human Services, protected all data from subpoena by federal, state, and local courts. Adolescents completed a battery of questionnaires and participated in a dyadic conflict discussion task with their fathers at age 13. Adolescents completed the AAI at age 14 and again at age 24. Participants were compensated and, when necessary, transportation and childcare were provided to participating families.

Measures

Attachment state of mind (ages 14 and 24).

The AAI (George et al., 1996) and Q-set (Kobak et al., 1993) were used to analyze adolescents’ descriptions of their relationships with both of their parents. The interview included 18 questions and took an average of one hour to complete. The original AAI is considered the gold standard for assessing attachment state of mind in adulthood (Hesse, 2016); in this study, questions were adapted slightly to be more natural and easily understood by adolescents. Interviews were audio-recorded and transcribed verbatim for coding.

The AAI Q-set (Kobak et al., 1993) was designed to parallel the AAI classification system (Main & Goldwyn, 1998); however, it also yields continuous dimensions of attachment organization, in line with current recommendations to measure attachment security on the AAI continuously (e.g., Roisman, Fraley, & Belsky, 2007). Final AAI scores take into account participants’ narratives about both parents. The data produced by the system can be reduced via an algorithm to classifications that have been found to largely agree with three-category ratings from the AAI Classification System, both in the field generally and when applied to a subsample of the present sample using coders from this laboratory (e.g., Allen et al., 2004). Coders read each transcript and provided a Q-set description by sorting 100 items into nine categories ranging from most to least characteristic of the interview, using a forced distribution. To establish validity, these Q-sets were then compared with dimensional prototypes for security (characterized by narrative coherence, emotional balance, and valuing close relationships), preoccupation (characterized by preoccupied anger or emotional overwhelm), and dismissing attachment (characterized by idealization or derogation of parents, memory failures, and downplaying the importance of close relationships) (see Kobak et al., 1993). The correlation of the 100 items of an individual’s Q-sort with each dimension (ranging on an absolute scale from −1.00 to 1.00) was then taken as the participant’s scale score for that dimension.

All interviews were independently rated by at least two coders with extensive training in both the Q-sort and the AAI classification system; the Spearman–Brown reliability for the overall security scale score was .82. Prior research in adolescent samples has indicated that security shows strong predictive validity and is highly stable over a 2-year period (r = .61; Allen et al., 2004). Inter-rater reliability in the present sample was calculated via the intraclass correlation coefficient (ICC); the ICC for the final security scale score was .82, in the excellent range for this statistic (Cicchetti & Sparrow, 1981).

Father acceptance and psychological control.

The Children’s Report of Parent Behavior Inventory (CRPBI; Schaefer, 1965) was used to measure adolescents’ perceptions of fathers’ acceptance and psychological control. The Parental Acceptance versus Rejection subscale contains 10 items measuring adolescents’ perceptions of parents as being affectionate, emotionally supportive, and egalitarian. Sample items include “My father figure cheers me up when I am sad” and “My father figure believes in showing his love for me.” The Psychological Control versus Psychological Autonomy subscale contains 10 items assessing the degree to which adolescents perceive their parents to use love withdrawal, guilt, anxiety, or other psychological methods to control their behavior. Example items include, “My father figure is a person who if I have hurt his feelings, stops talking to me until I please him again,” “My father figure is a person who wants to control whatever I do,” and “My father figure is a person who is always trying to change me.” Adolescents responded on a three-point Likert scale (Not like [my father]; Somewhat like [my father]; A lot like [my father]). Final scores for each subscale were obtained by summing responses on the frequency scale across behaviors and could range from 10 to 30. Cronbach’s alphas for each subscale were strong (Acceptance vs. Rejection = .89, Psychological Control vs. Autonomy = .82).

Observed autonomy and relatedness with dad (age 13).

Adolescents and their fathers participated in a conflict task in which they discussed an issue in their relationship that they had separately identified as an area of disagreement. Typical topics include money, grades, household rules, friends, sibling issues, and division of household tasks. Interactions lasted 8 minutes and were videotaped, transcribed, and coded with the Autonomy and Relatedness Coding System (Allen et al., 2000). The present investigation focuses on two dimensions of father behavior toward the teen: (a) positive relatedness, including validating statements and displays of engagement and empathy with the teen; and (b) autonomy support, including providing reasons behind one’s position and maintaining a calm, confident tone in the discussion without shutting down the teen’s point of view. Reliability for these scales was in the excellent range (ICCs = .83, .86, respectively).

Father aggression.

The Parent-Child version of the Conflict Tactics Scales (CTS; Straus, 1979; Straus et al., 1998) contains 80 items assessing severity and frequency of conflict between two partners, as well as conflict management techniques. The 6-item Psychological Aggression subscale and the 11-item Physical Aggression subscale were used in the current study. Adolescent reports of their father’s psychological and physical aggression were used, given that adolescent perceptions of parental behavior are often equal to or exceed the accuracy of parental reports and are an important predictor of outcomes (Albrecht et al., 2007; Donovan & Brassard, 2011). Sample items from the Psychological Aggression subscale include “Your father insulted or swore at you,” “Your father did or said something to spite you,” and “Your father threatened to hit or throw something at you.” Sample items for the Physical Aggression subscale include “Your father pushed, grabbed, or shoved you,” “Your father beat you up,” and “Your father hit or tried to hit you with a belt, hairbrush, paddle, stick, or similar item.” Adolescents responded on a seven-point Likert scale (never; once; twice; 3–5 times; 6–10 times; 11–20 times; or more than 20 times). Final scores were obtained by summing responses on the frequency scale across behaviors and could range from 0 to 36 for the Psychological Aggression subscale and 0 to 66 for the Physical Aggression subscale. Cronbach’s alphas were .72 for the Psychological Aggression subscale and .55 for the Physical Aggression subscale, likely due to the restricted range in the present sample (see below for how the physical aggression data were handled).

Attrition Analyses

Of the original 184 participating families, 174 adolescents (95%) provided attachment data at age 14, and 157 (85%) provided attachment data ten years later, at age 24. At the initial assessment, 158 teens (86%) were able to report on their fathers’ behavior, and 109 fathers (59%) reported on their own parenting. Attrition analyses showed that boys were more likely to be missing the age 24 AAI (22%) compared to girls (8%), χ² = 7.44, p = .006; no other baseline characteristics were significantly different.

Observations of dyadic behavior were available for 97 father–teen dyads. Adolescents whose fathers participated in the initial observational assessment were more likely to be boys (64%) than girls (44%), χ² = 5.88, p = .015; were more likely to be living with their father (75%) vs. not (28%), χ² =39.98, p < .001; and were from higher-income households, on average, t = 5.65, p < .001. However, adolescents whose fathers did vs. did not participate in the initial observational assessment did not differ on any teen-reported baseline variables (acceptance, aggression, psychological control).

Analytic Plan

Preliminary analyses examined distributional properties of the data and bivariate correlations among all variables of interest. Although attachment security was the focus of the present hypotheses, for completeness we report correlational findings with the dimensions of insecurity (dismissing and preoccupied attachment).

To test focal hypotheses, separate regression models predicting attachment security at age 14 and 24 were run in MPlus, using full information maximum likelihood (FIML) to account for missing data. To mitigate potential biases in results, adolescent gender and family income (calculated as a percentage of the federal poverty line at the initial assessment) were included as a priori covariates in an initial step in all models. To examine promotive factors predicting attachment security (Hypotheses 1 and 2), father residential status was entered in Step 2, and autonomy/relatedness variables in Step 3. To examine threat-related risk factors predicting insecurity (Hypothesis 3), covariates were entered in Step 1, and fathers’ physical and verbal aggression were entered together in Step 2. To test the contributions of fathers over and above mothers (Hypothesis 4), mother versions of each predictor were entered into the above models. Tests of multicollinearity confirmed that all models had variance inflation factors below 5.0. Predictors were allowed to covary.

Results

Preliminary Analyses

Descriptive statistics and bivariate correlations are summarized in Table 1. Teen reports of father physical aggression were positively skewed (skewness = 3.11); thus, this variable was dichotomized for analyses, such that 0 = no physical aggression and 1 = any physical aggression. Overall, n = 30 adolescents (20%) reported that their fathers used any physical aggression toward them during conflict.

Table 1.

Descriptive Statistics and Bivariate Correlations Among Study Variables.

Variable name	N	M	SD	1	2	3	4	5	6	7	8	9	10	11	12
1. Father Residential Status (13y)	184	-	-	-
2. Father Obs. Autonomy Support (13y)	97	3.05	0.60	−.04	-
3. Father Obs. Pos. Relatedness (13y)	97	2.15	0.71	.26*	−.07	-
4. Father Acceptance v. Rejection (13y)	156	24.50	4.47	.19*	.08	.20*	-
5. Father Psych. Control v. Autonomy (13y)	158	14.43	3.80	−.16	.03	−.27**	−.25**	-
6. Father Verbal Aggression (13y)	158	3.74	4.65	−.06	.17	−.07	−.40**	.34**	-
7. Father Physical Aggression (13y)	150	0.55	1.42	−.03	.08	−.18	−.31**	.23**	.50**	-
8. TN Attachment Security (14y)	174	0.25	0.42	.34**	.06	.30**	.16*	−.26**	−.12	−.31**	-
9. TN Dismissing Attachment (14y)	174	0.05	0.43	−.42**	−.08	−.28**	−.20*	.26**	.11	.25**	−.94**	-
10. TN Preoccupied Attachment (14y)	174	-0.03	0.23	−.25**	.00	−.17	−.23**	.19*	.26**	.33**	−.55**	.46**	-
11. TN Attachment Security (24y)	157	0.03	0.46	.18*	−.10	.13	.16	−.21*	−.26**	−.21*	.46**	−.47**	−.31**	-
12. TN Dismissing Attachment (24y)	157	0.23	0.43	−.15	−.05	−.16	−.16	.25**	.20*	.11	−.43**	.47**	.16	−.85**	-
13. TN Preoccupied Attachment (24y)	157	0.12	0.34	−.15	.12	−.04	−.10	.12	.23**	.19*	−.20*	.16	.47**	−.55**	.29**

Note. TN = target teen, Obs. = observed, Pos. = positive, Psych. = psychological; residential status is coded such that 0 = teen not living with father, 1 = teen living with father; ages of assessment are shown in parentheses.

^*p < .05,

^**p < .01.

Father variables that showed significant associations with attachment security in adolescence included living in the same household with the father, fathers’ observed positive relatedness toward teens during dyadic discussions, acceptance of the teen, lower use of psychological control, and lack of physical aggression toward the teen during conflict (see Table 1). Dismissing attachment showed inverse associations with each of these variables. Preoccupation was associated with teens’ perceived rejection by their father and fathers’ use of psychological control, verbal aggression, and physical aggression.

As shown in Table 1, father variables that showed significant associations with attachment security in adulthood included living in the same household with the father, fathers’ lower use of psychological control, lower verbal aggression, and lack of physical aggression at age 13. Dismissing attachment showed positive associations with each of these variables. Preoccupation in adulthood was associated with fathers’ use of verbal and physical aggression at age 13.

Initial independent samples t-tests were run to examine differences in attachment by father residential status (see Figure 1A). Adolescents who lived with their biological or adoptive father at age 13 showed significantly higher age 14 attachment security (t = 4.69, p < .001, Cohen’s d = .72), less dismissing attachment (t = −6.03, p < .001, Cohen’s d = −.92), and less preoccupation (t = −3.34, p < .001, Cohen’s d = −.51), compared to those who did not. Those who lived with their father at age 13 also showed significantly higher attachment security even at age 24 (t = 2.25, p = .026, Cohen’s d = .36), compared to those who did not live with their fathers, though differences in dimensions of insecurity were no longer significant (ps >.06).

Figure 1.

Simple and interactive effects of father residential status predicting adolescent attachment security on the Adult Attachment Interview. A) Mean scores for adolescent attachment security at ages 14 and 24, depicted by father residential status. B) Interactions of father acceptance and residential status predicting adolescents’ attachment on the Adult Attachment Interview at age 14 and C) at age 24.

Note. Age of assessment is in parentheses.

*p < .05.

Hypothesis Testing

Results of the regression models for father promotive factors are shown in Table 2, and results for father threat-related risk factors are shown in Table 3. Regarding covariates, girls and adolescents with higher family income had higher attachment security scores, on average, at age 14 and at age 24. Results for all focal predictors (after accounting for covariates) are described below.

Table 2.

Stepwise regression models of fathers’ secure base support at age 13 predicting adolescent attachment security at ages 14 (left columns) and 24 (right columns).

	AAI Attachment Security Age 14				AAI Attachment Security Age 24
	β	p	ΔR2		β	p	ΔR2
Step I. Covariates			.09				.07
TN gender	.16	.025			.23	.002
Family income	.28	<.001			.17	.023
Step II. Residential Status			.07				.01
TN living with father	.28	<.001			.11	.175
Step III. Autonomy & Relatedness			.10				.06
Father acceptance v. rejection	.03	.683			.07	.408
Father psychological control v. autonomy	−.17	.037			−.15	.083
Autonomy support (Obs.)	.05	.578			−.07	.499
Positive relatedness (Obs.)	.25	.015			.13	.293

Note. Predictors were allowed to covary. TN = teen, Obs. = observed; physical aggression is coded such that 0 = no physical aggression, 1 = any physical aggression. boldface indicates significant parameters, p < .05.

Table 3.

Regression models of father aggression toward the teen at age 13 predicting adolescent attachment security at ages 14 and 24.

	AAI Attachment Security Age 14				AAI Attachment Security Age 24
	β	p	ΔR2		β	p	ΔR2
Step I. Covariates			.09				.07
TN gender	.16	.025			.23	.002
Family income	.28	<.001			.17	.023
Step II. Threat			.07				.05
Father verbal aggression	.04	.640			−.22	.016
Father physical aggression	−.29	.001			−.02	.867

Note. Predictors were allowed to covary. Physical aggression is coded such that 0 = no physical aggression, 1 = any physical aggression; TN = teen; boldface indicates significant parameters, p < .05.

Hypothesis 1: Father residential status.

Father residential status was dummy-coded for analysis (0 = teen not living with biological or adoptive father at age 13; 1 = teen living with father). There was a significant main effect of fathers’ residential status at age 13 predicting attachment security at age 14, accounting for an additional 7% of the variance in adolescent attachment security over and above gender and family income. However, this effect was non-significant at age 24 after accounting for covariates (see Table 2).

To rule out potential socioeconomic confounds, we ran a follow-up model with race and fathers’ education level as additional covariates. Even after accounting for adolescent gender, family income, paternal education level, employment status, and the number of people the father was supporting financially, father residential status at age 13 remained a significant predictor of adolescent attachment security at age 14 (b = .21, SE = .09, p = .015).

To examine potential developmental effects, exploratory analyses examined father residential status at a later time point: age 16. Regarding predictive effects on adult attachment, father residential status at age 16 significantly predicted attachment security at age 24 (b = .16, SE = .07, p = .025), accounting for an additional 3% of the variance in adult attachment beyond gender and income. Regarding stability of father residential status, n = 78 adolescents were living with their father consistently from age 13–16 (stable-residential group), n = 31 experienced changes in father residential status from age 13–16 (unstable group), and n = 75 adolescents were not living with their father at either age (stable-nonresidential group). These three groups differed with respect to their attachment security at age 24, F(2, 154) = 5.83, p = .004, η² = .07. Bonferroni-corrected post-hoc tests showed that teens who lived with their fathers consistently across age 13–16 had higher levels of attachment security at age 24 compared to both the stable-nonresidential group (p = .019) and the unstable group (p = .015), whereas the stable-nonresidential and unstable groups did not differ from one another.

Hypothesis 2: Father support for autonomy and relatedness with teen.

Autonomy and relatedness variables were entered together in Step 3, accounting for covariates and father residential status. Teens’ perceptions of fathers’ psychological control vs. autonomy predicted lower attachment security in adolescence; perceptions of fathers’ acceptance vs rejection were not significant. In observed father–teen interactions at age 13, fathers’ positive relatedness (but not autonomy support) predicted secure state of mind regarding attachment at age 14. When attachment was assessed in adulthood, however, these pathways were no longer significant (see Table 2).

Hypothesis 3: Father aggression as a threat to security.

Teen perceptions of fathers’ physical aggression during conflict at age 13 predicted lower attachment security at age 14 (accounting for verbal aggression), whereas fathers’ use of verbal aggression at age 13 predicted attachment insecurity at age 24 (Table 3).

Hypothesis 4: Unique contributions of fathers and mothers.

Main effects of mother predictors have been reported elsewhere (see Allen et al., 2003); here, we focus on the unique effects of father predictors. When mother versions of the variables were included in each of the above models, the effects of fathers’ psychological control and positive relatedness were no longer significant (ps > .05). However, father residential status (b = .21, SE = .06, p = .001) and physical aggression (b = −.27, SE = .10, p = .006) uniquely predicted attachment security at age 14, over and above effects of mother residential status and physical aggression, respectively. Interestingly, mothers’ physical aggression — though rare (n = 40 teens reported any use of physical aggression by mothers at age 13) — uniquely predicted attachment at age 24 (b = −.23, SE = .10, p = .015).

Exploratory Analyses

To examine interactions between father and mother predictors, their two-way interaction terms were entered as a third step in the models for Hypothesis 4. Continuous variables were standardized prior to creating interaction terms. There were no significant interactions between father and mother variables predicting attachment in adolescence or adulthood, all ps > .05.

When all father variables were entered simultaneously into the model predicting attachment at age 14, father residential status (β = .17, p = .011) and physical aggression (β = −.25, p = .007) had significant unique effects (model R² = .30). When all father variables were entered into the model predicting attachment at age 24, no variables showed significant unique effects; however, girls continued to show higher security scores than boys, on average (β = .22, p = .006; model R² = .16).

To examine moderation of developmental pathways by adolescent and father characteristics, two-way interactions between each focal predictor and moderator were entered as an additional step in the models for Hypotheses 2–3. Continuous variables were standardized prior to creating interaction terms. Interactions were probed using simple slopes analysis, and significant results are described below.

Moderation by Father Residential Status.

Father residential status at age 13 moderated the effect of father acceptance on attachment at age 14 (b = −.21, SE = .07, p = .002; Figure 1B) and at age 24 (b = −.16, SE = .08, p = .044; Figure 1C). Specifically, teens’ perceptions of father acceptance vs. rejection at age 13 predicted attachment security at age 14 for teens living with their fathers (b = .12, SE = .05, p = .007), but not for those with non-residential fathers (b = −.08, SE = .05, p = .086). Similarly, father acceptance predicted attachment security at age 24 for teens who lived with their fathers (b = .11, SE = .06, p = .054), but not for those with non-residential fathers (b = −.04, SE = .06, p = .438). No other links differed by father residential status, ps > .05.

Moderation by Adolescent Gender.

A significant interaction emerged between adolescent gender and father acceptance predicting attachment security at age 14 (b = −.14, SE = .07, p = .034), such that the observed link was stronger for girls (b = .08, SE = .04, p = .056) than for boys (b = −.05, SE = .05, p = .304). Adolescent gender interacted with fathers’ verbal aggression to predict attachment security at age 14 (b = −.14, SE = .07, p = .039), such that the link was significant for girls (b = −.11, SE = .05, p = .022), but not for boys (b = .03, SE = .05, p = .483). No other pathways differed by gender.

Moderation by Adolescent Race.

Pathways linking father psychological control, positive autonomy and relatedness, and verbal aggression to attachment at age 14 varied by adolescents’ racial identity (coded 1 = Black, 0 = non-Black). Specifically, there was an interaction between race and psychological control (b = −.28, SE = .09, p = .001), such that fathers’ psychological control was associated with lower attachment security at age 14 specifically for Black adolescents (b = −.29, SE = .07, p < .001), but not for non-Black adolescents (b = −.01, SE = .04, p = .865). Race also interacted with observed autonomy support (b = .26, SE = .13, p = .043) and positive relatedness (b = −.21, SE = .10, p = .041) to predict attachment security at age 14, such that autonomy support was a significant predictor for Black adolescents (b = .31, SE = .12, p = .011), but not non-Black adolescents (b = .01, SE = 04, p = .863), whereas positive relatedness was a significant predictor for non-Black adolescents (b = .12, SE = .05, p = .013). Additionally, race moderated the link between fathers’ verbal aggression and attachment security in adulthood (b = .18, SE = .08, p = .033), such that the link was significant among non-Black adolescents (b = −.17, SE = .05, p = .001), but not Black adolescents (b = .01, SE = .07, p = .856). No other pathways varied by racial identity.

Moderation by Family Income.

Family income interacted with teens’ perceptions of fathers’ acceptance vs. rejection to predict attachment at age 24 (b = .07, SE = .04, p = .046); specifically, father acceptance predicted adult attachment security among high-income families (+1 SD above the mean; b = .12, SE = .06, p = .036), but not low-income families (−1 SD below the mean; b = −.03, SE = .05, p = .569). No other pathways differed by family income.

Discussion

The present study illuminated father characteristics and behaviors in early adolescence that contributed to teens’ attachment security at age 14 and ten years later, at age 24. The most consistent finding to emerge was the central role of father residential status—both as a unique predictor of adolescent attachment security and as a moderator of other factors. Beyond fathers’ presence in the home, fathers’ observed positive relatedness toward the teen and adolescents’ perceptions of fathers’ support for their psychological autonomy (vs. control) predicted attachment security in adolescence. Fathers’ physical aggression predicted lower attachment security in adolescence, whereas fathers’ verbal aggression uniquely predicted lower security in adulthood, suggesting potential developmental shifts. Critically, fathers’ residential status and physical aggression showed links to attachment above and beyond the contributions of mothers. Moderation analyses revealed that multiple pathways varied by family income, adolescent gender, and racial identity, underscoring the importance of examining diversity in father–adolescent relationships. Broadly, dimensions of paternal care assessed at age 13 were stronger predictors of attachment the following year (age 14), but effects appeared to weaken by early adulthood (age 24). Findings highlight fathers’ contributions to secure attachment representations during a sensitive period of social development.

Residential Status: The Importance of Fathers’ Presence

Initial analyses revealed that teens who did not live with their father showed significantly lower attachment security in mid-adolescence, replicating previous cross-sectional findings that teens report less secure attachment to nonresidential fathers (Furstenberg & Harris, 1993; Williams & Kelly, 2005). Extending previous work, the present study yielded several additional insights: First, the effect of residential status on adolescent attachment was unique to fathers, remaining significant even when mothers were included in analyses (although there was much less variation in mother residential status in our sample). Second, father residential status was a surprisingly robust predictor of attachment security in adolescence, showing a large effect size and remaining significant above and beyond family income and all other measures of parenting quality assessed in the present study. Third, residential status emerged as an important moderator of other paternal behaviors (discussed below), in line with research suggesting that the consequences of fathers’ parenting quality may depend on their presence in teens’ lives (Brown et al., 2007).

Developmentally, the magnitude of the difference in attachment security by father residential status was substantial in mid-adolescence but faded as teens entered early adulthood. It may be that fathers’ physical proximity is more important earlier in development before teens fully differentiate from parents, but that other factors such as peer availability and friendship quality become more important for attachment-related outcomes over time (e.g., Allen & Miga, 2010; Allen et al., 2021). Similarly, having one’s father living in the family home may be less important for attachment after adolescents leave the home in early adulthood. However, exploratory analyses suggested that fathers’ residential status at age 16, as well as the stability of father residential status from age 13–16, predicted attachment in early adulthood, though effect sizes were small. We encourage future work on developmental timing to better understand these pathways.

Father Secure Base Provision in Adolescence: The Importance of Autonomy and Connection

Extending foundational work on fathers’ sensitive and challenging play as forms of secure base provision in early childhood (Grossmann et al., 2002), the present study suggests that low psychological control and support for autonomy and relatedness may be markers of paternal secure base provision during adolescence. In the domain of autonomy support, fathers who supported adolescents’ psychological autonomy and refrained from psychological control had adolescents who were more likely to be securely attached in adolescence. Thus, paternal care that attempts to manipulate the teen’s behavior or opinions may undermine adolescents’ need for autonomy in attachment relationships. Findings are consistent with hypotheses and previous work linking self-reported attachment insecurity to experiences of psychological control from parents generally (Cai et al., 2014; Choe et al., 2020) and fathers specifically (Leondari & Kiosseoglou, 2002). Although fathers’ psychological control was significantly correlated with lower attachment insecurity at age 24, this link was no longer significant when other security-promoting factors were included in the final model. Pathways were not moderated by father residential status, adolescent gender, or family income, but differed by racial identity—specifically, Black youth were particularly impacted by fathers’ psychological control. Previous work has linked fathers’ psychological control to related outcomes such as self-esteem among both Black and White teens (Bean et al., 2003); thus, findings warrant replication and extension to other racial-ethnic groups.

Contrary to predictions, fathers’ observed autonomy support during dyadic interactions was unrelated to attachment security in the overall sample, whereas fathers’ positive relatedness was associated with greater attachment security at age 14. Interestingly, however, findings varied by adolescent racial identity: Fathers’ autonomy support predicted adolescent attachment security specifically for Black teens, whereas positive relatedness was more important for non-Black teens. Results are broadly consistent with the notion that parental secure base provision may vary as a function of children’s social context (Dunbar et al., 2017, 2022; Stern et al., 2022). It is possible that fathers’ support for autonomy is particularly important for Black teens’ sense of confidence in a secure base while navigating racism-related stressors in the wider world.

Teens’ perceptions of being accepted by their father at age 13 were correlated with greater attachment security at age 14, but the role of paternal acceptance varied by gender, father residential status, and family income. Specifically, fathers’ acceptance vs. rejection toward teens when they were 13 was associated with greater attachment security for girls, teens who were living with their father, and those from high-income families, suggesting that feeling accepted by one’s father may be more important for teens with higher levels of privilege. For these groups of teens, the positive associations between paternal acceptance and attachment security align with parental acceptance–rejection theory (PARTheory; Rohner et al., 2012), which postulates that experiences of parental acceptance are foundational for the children’s sense of security and positive self-concept, with long-term consequences for adult adjustment. From an attachment perspective, experiences of parental rejection are thought to contribute to insecure attachment and specifically avoidance (Ainsworth et al., 1978); indeed, in the present study, perceptions of paternal rejection (i.e., lower acceptance) were correlated with insecure attachment, and most strongly with dismissing attachment. Previous studies have shown concurrent associations between child-reported paternal acceptance and secure attachment to fathers on the Security Scale in school-aged children (Ildiz & Ayhan, 2022).

Threats to a Safe Haven in Adolescence: Father Physical and Verbal Aggression

In addition to support for autonomy and exploration, adolescents also turn to parents for protection— that is, a safe haven (Rosenthal & Kobak, 2010). Thus, if a parent becomes a source of threat rather than safety, it may threaten teens’ sense of security in close relationships. Building on previous findings linking fathers’ hostility to teen insecurity (Allen et al., 2007), we sought to disentangle physical and verbal dimensions of fathers’ aggression during conflict. In the present investigation, fathers’ physical aggression was a particularly robust predictor of attachment insecurity at age 14, remaining significant even when all other father variables were included in the model, and contributing unique variance over and above mothers’ physical aggression. Moreover, effects of physical aggression were not moderated by teen or family characteristics, suggesting that fathers’ physical aggression is equally detrimental to attachment security across diverse groups of adolescents. Findings align with the broader literature linking childhood experiences of parental physical abuse to insecure attachment (e.g., Murphy et al., 2014).

Extending this work, whereas teen reports of fathers’ physical aggression during conflict at age 13 uniquely predicted lower attachment security at age 14 (accounting for verbal aggression), fathers’ use of verbal aggression at age 13 uniquely predicted attachment insecurity at age 24. We interpret these findings to suggest that experiences of physical aggression may be more threatening and dysregulating when teens are younger and more dependent on parents to provide a safe haven, but may become less salient over time. Some data suggest that recent experiences of parental physical abuse are more strongly related to adolescent attachment, whereas physical abuse five years earlier was not (Sternberg et al., 2005). As teens leave home and gain independence in early adulthood, it may be the emotional wounds of verbal aggression that have more lasting influence on their internal working models of the self and others. Interestingly, correlational results suggest that father verbal aggression may be most strongly associated with the preoccupied dimension of attachment insecurity, possibly reflecting teens’ anger regarding past experiences of being the target of verbal aggression.

Notably, effects of verbal aggression varied by teen gender and race, suggesting that girls may be more impacted in early adolescence, and Black youth may be less impacted by early adulthood. It is possible that girls show the negative impacts of fathers’ verbal aggression earlier in development, but that this gender difference narrows by adulthood. Given that levels of father aggression were modest in the present sample, is also possible that occasional father verbal aggression, within a broader context of high parental support and other promotive factors such as positive racial-ethnic socialization, is less detrimental to attachment security among Black adolescents (for similar ideas see Dunbar et al., 2017, 2022). Nevertheless, parental verbal aggression has been shown to have negative consequences for other aspects of adjustment across gender and racial groups, with research demonstrating links to Black boys’ self-control and behavioral outcomes (e.g., Evans et al., 2012).

Strengths, Limitations, and Future Directions

Strengths of the study include the long-term longitudinal design to examine developmental pathways across ages 13 to 24, the use of multimethod assessment of father–teen relationships (including observational measures), assessment of attachment representations using the gold standard Adult Attachment Interview at two time points spanning a decade, and the focus on fathers in understanding individual differences in adolescent attachment— taking into account key sources of variation such as race and gender. Crucially, findings highlight key unique contribution of paternal care in adolescence, above and beyond well-established effects of maternal care and socioeconomic factors.

Nevertheless, findings should be contextualized in light of study limitations. First, father residential status provides only a crude measure of fathers’ presence in teens’ lives. We did not capture shared custody arrangements, variation in time spent with fathers, proximity to non-residential fathers’ home, duration of the living arrangement, the presence of other father figures, or reasons for fathers’ non-residential status (however, over half of non-residential fathers were separated or divorced from the teen’s mother). Thus, it is unclear whether effects are driven by father residential status per se, versus experiences of divorce, father abandonment, incarceration, or short-term parental separation. Future research should examine factors that may alter the meaning of fathers’ residential status for the child, such as when and over how long a period non-residence occurred, the reasons it occurred, the ongoing relations with the biological mother, and the nature and extent of ongoing contact with the child. In addition, we emphasize that ultimately, it is the quality of the father–teen interaction that matters for attachment security, more than the quantity of interaction; thus, future work should continue to examine the quality of father’ presence and involvement, including contributions of high-quality caregiving by non-residential fathers.

Second, for the observational measures, we could only include fathers who were sufficiently involved to be willing to attend the laboratory session with their teen. Third, although the longitudinal design allows for directional pathways from father predictors to later adolescent outcomes, we cannot draw causal conclusions from these correlational data. It is possible that other factors—such as marital conflict, fathers’ own attachment, household chaos, or external family stressors—may influence both fathers’ behavior and adolescents’ attachment representations. Additionally, we did not examine predictors of relative change in attachment over time (but see Allen et al., 2004, 2018) or test for significant differences between pathways predicting attachment at ages 14 vs. 24, so we interpret any differences with caution; additionally, due to the modifications made to the AAI for developmental appropriateness with adolescents, security scores between the two age groups are not directly comparable.

Importantly, the Q-sort scoring of the AAI did not permit examination of “probable childhood experiences with mother/father” or parent-specific state of mind scores (e.g., idealization, anger). On the other hand, it is notable that independent observations and questionnaire measures of adolescent experiences with each parent prospectively predicted generalized state of mind scores, suggesting that experiences with fathers inform attachment state of mind more broadly, beyond representations of the father specifically. Nevertheless, future research should investigate the extent to which fathers’ residential status and aggression relate to father-specific experiences and states of mind on the AAI.

Finally, although the sample was representative of the community from which it was drawn, we did not have sufficient sample size to examine all racial identities. Critically, the present sample allowed for new insight into pathways unique to Black adolescents—an important and understudied population in the attachment literature (Murry et al., 2022; Stern et al., 2023). However, the “non-Black” category (used to preserve statistical power) collapsed important variation in the sample, which also included small numbers of Asian, Latino/a, and multiracial teens. Given our sample size, we regard all moderation analyses as exploratory and strongly encourage replication. Future work in larger samples should continue to examine attachment relationships among fathers and adolescents of different racial identities, as well as potential intersections between fathers’ caregiving, adolescent gender and race. Additionally, future research should examine culturally specific forms of father secure base provision, such as positive racial-ethnic socialization (Dunbar et al., 2022; Tyrell & Masten, 2022; Stern et al., 2023).

Conclusions and Implications

The present investigation provides the most comprehensive examination to date of fathers’ contributions to secure attachment representations in adolescence. In summary, teens who lived with their fathers, experienced psychological autonomy and positive relatedness, and reported low levels of paternal aggression were more likely to be securely attached in adolescence and, to a lesser extent, in early adulthood. Findings underscore the importance of fathers’ presence, support for teens’ autonomy, and non-aggression in shaping short- and long-term outcomes for adolescents’ state of mind in close relationships. Importantly, developmental pathways were moderated by father residential status, adolescent gender and race, and family income.

The present findings contribute to a growing literature on “father love” (Hrdy, 2024; Li & Meier, 2017; Rohner & Veneziano, 2001) and have key implications for practice: Family therapy and attachment-focused interventions for adolescents should involve fathers (including non-residential fathers) whenever possible, recognizing their unique contributions to adolescent social-emotional development. For divorced or separated families, providing opportunities for fathers to stay present in teens’ lives, and providing support for both residential and non-residential fathers to develop sensitive caregiving competencies, may be especially important for adolescents’ attachment. In particular, practitioners can focus on (a) increasing fathers’ capacity to be physically and emotionally present, (b) reducing threatening behaviors such as physical and verbal aggression (for example, through strengthening emotion regulation and communication skills), and (c) supporting healthy father–teen conflict resolution (through practicing constructive patterns of relationship rupture-and-repair that preserve teens’ autonomy and relatedness with the father and reduce psychologically controlling behaviors). For parents themselves, our findings underscore that fathers’ quality of care and presence in teens’ lives matters—both in the short term and into the early adult years—and fathers who stay present and supportive even as teens gain more autonomy play a key role in building adolescents’ sense of security in their close relationships.

Data availability statement:

Data are not publicly available, due to ethical/ confidentiality restrictions.