Early relationships of adolescents

Abstract

The purpose of the study was to examine the relationship between attachment behavior, substance use, and sexual activity among adolescents in romantic relationships. Concepts from attachment theory have been proposed to explain adolescent risk behavior; however, research on adolescent attachment rarely assesses both members of the couple. A total of 21 adolescent couples (aged 15–18 years) were assessed on self-report and observed measures of romantic attachment, substance use, and sexual activity. Attachment anxiety was related to one’s own risk behavior. Self-reported attachment avoidance was also related to one’s own and his/her partner’s observed attachment behaviors. Given these results, risk prevention/intervention programs targeted toward adolescents may benefit from assessing attachment representations of adolescents, which may be related to participation in risk behavior.

Keywords

Adolescents risk behavior romantic attachment romantic relationships

Establishing and navigating early romantic relationships is a pivotal task during adolescence, but only recently has the scientific community begun to examine adolescents’ early relationship experiences (Collins, Welsh, & Furman, 2009). While once deemed trivial and inconsequential, adolescent romantic relationships are now increasingly believed to have significant long-term effects on later adult relationships (Karney, Beckett, Collins, & Shaw, 2007). During adolescence, the majority of youth experience their first romantic relationship. These relationships typically become a central focus of the youth’s life and can be associated with the initiation of risk behaviors such as substance use (Eaves, 2004) and sexual exploration (Zimmer-Gembeck, Siebenbruner, & Collins, 2004).

Substance use and risky sexual activity are especially salient issues in adolescence as 7.6 million of 12- to 17-year olds reported using alcohol and 4.7 million reported illicit drug use in the past year in the US (Office of Applied Studies, 2007). Additionally, adolescents account for almost half of the sexually transmitted infection cases reported each year (Centers for Disease Control, 2012). Current intervention efforts have shown some success in reducing risk behavior among adolescents. Programs utilizing the family as a source of support are associated with reduced substance use and delinquent behavior (Feldstein & Miller, 2006). Yet, the inclusion of the family in treatment may not be beneficial or safe for some adolescents due to past or current child abuse, substance use, or mental illness among parents. Therefore, new strategies, such as exploring the influence of a romantic partner on treatment, should be investigated. Furthermore, exploring the relationship between romantic attachment and risk behavior may aid in the development of targeted and specific prevention and intervention strategies for service providers.

Attachment theory

Attachment theory is one of the leading theories in the study of adolescent romantic relationships (Collins et al., 2009). According to Bowlby (1973), an individual develops a specific view of his or her world based on repeated experiences with a caregiver. The repeated interactions become internalized to create an internal working model that shapes the individual’s view of himself/herself and also his or her view of others (Bowlby, 1969). Secure attachment is related to a positive view of the self and others resulting in generally satisfying and long-term relationships (Cooper, Shaver, & Collins, 1998). Individuals with a negative view of the self, but positive view of others, tend to report high attachment anxiety in which they are unsure of their caregiver or partner’s commitment leading to a fear of rejection or abandonment (Reese-Weber & Marchand, 2002). Individuals with high attachment anxiety consistently seek more intimacy and security in their relationships with others (Hazan & Shaver, 1987). Individuals with avoidant attachment orientations have a positive view of the self and negative view of others. Consequently, they tend to be self-reliant and cope with negative emotions by withdrawing from others during times of distress (Cooper et al., 1998).

Adolescent attachment

Attachment orientations are considered to be relatively stable over the life span, yet difficulties in measuring attachment among adolescents have been noted. For example, the use of self-report questionnaires is common in attachment research, and although valid and reliable self-report measures exist, most were created using adult samples. Whether these measures are appropriate for a younger sample continues to be debated with researchers advocating for further investigation into the use of current adult attachment measures with adolescents (Weinfield, Whaley, & Egeland, 2004). Additionally, collecting data from multiple perspectives is preferable in couple research (Collins et al., 2009), yet few studies have been successful in collecting data directly from peers or romantic partners (Campbell, Simpson, Kashy, & Rholes, 2001). Current research related to attachment and adolescent risk behavior at the individual level suggests that when combined with an insecure attachment (i.e., high attachment anxiety or avoidance), the incidence of substance use and sexual activity may be higher in comparison to youth with secure attachment orientations (Kassel, Wardle, & Roberts, 2007; Strachman & Impett, 2009).

Attachment and adolescent risk behavior

Sexual activity

A natural progression of the study of romantic attachment in adolescence involves an analysis of sexual activity. Indeed, adolescence is a time of sexual exploration with between 28.4 and 56.2% of heterosexual adolescents in grades 9–12 reporting sexual intercourse (Centers for Disease Control and Prevention, 2011). Studies indicate that avoidant adolescents are more likely than anxious adolescents to report having sex to fit in with peers, brag about sexual activity to their friends, or lose their virginity (Schachner & Shaver, 2004). Anxious individuals, however, indicated that sexual motivation emerged from a need to feel closer to their partner or to make their partner love them more. Similar results were discovered by Tracy, Shaver, Albino, and Cooper (2003) with avoidant adolescents having sex in order to lose their virginity, and those with anxious attachments using sexual relations as a source of security in the relationship.

Social skills may also be linked to attachment and sexual activity. Avoidant adolescents have demonstrated limited social skills, less involvement in risk behavior, and little interaction with sexual partners (Cooper et al., 1998). Those avoidant individuals who are sexually active have shown less restrictive beliefs about casual sex, and also tend to have engaged in their first sexual experience before the age of 16 years (Gentzler & Kerns, 2004). Adolescents with anxious attachments demonstrate difficulties with social skills as well. In their attempt to satisfy their romantic partners, anxious adolescents often worried that asking a partner to use a condom would show the partner a lack of trust. Such low condom self-efficacy was related to higher rates of unprotected sex with risky partners (Kershaw et al., 2007). The ease of pressuring anxious individuals into risky sexual behavior (Gentzler & Kerns, 2004) is especially problematic when it occurs with an avoidant partner seeking power and control from the encounter (Schachner & Shaver, 2004).

Substance use

Similar to sexual activity, an individual’s first use of substances typically occurs during adolescence. Over half (55%) of adolescents report being intoxicated at least once by the time they reach 12th grade (Johnston, O’Malley, Bachman, & Schulenberg, 2008). Current research suggests that when combined with an insecure romantic attachment, the incidence of substance use may be higher as insecure adolescents have demonstrated more problematic substance use behaviors than secure adolescents. In particular, anxious individuals tend to report the most use in comparison to secure and avoidant youth (Cooper et al., 1998; Kassel et al., 2007). Consistent with the explanations given in response to the findings regarding sexual activity, researchers theorize that anxious individuals use substances as a way to cope with stress and engage with others (Cooper et al., 1998; Golder et al., 2005).

The association between avoidance and substance use is more complex. Some research demonstrates similar use patterns between avoidant and secure youth with avoidant individuals typically refraining from substance use due to a lack of social skills that inhibit their exposure to peer pressure (Cooper et al., 1998). Others have posited that avoidant individuals can take on the behavior characteristics of both secure and anxious individuals (Gentzler & Kerns, 2004; Schachner & Shaver, 2004). Those scoring high on attachment avoidance, especially adolescents, tend to avoid sex and drugs resulting in older ages at first use leading to risk patterns similar to secure persons in their early years. However, after first exposure to risk and into adulthood, rates of participation tend to increase and begin to mirror that of anxious individuals. This may be due to a more relaxed and accepting view of these behaviors by avoidant individuals (Tracy et al., 2003).

Actor and partner effects

Attachment anxiety and avoidance have been posited to explain the behaviors of individuals (Golder et al., 2005). That is, individuals with insecure attachments may increase their own substance use to cope with stress. The direct relationship between one’s attachment and his or her own behavior is known as an actor effect (Kenny, Kashy, & Cook, 2006). However, the influence of the individual’s characteristics on the behavior of his or her partner, known as a partner effect, is less frequently addressed in attachment research. One of the limited studies in this area indicated that individuals with avoidant partners, in comparison with secure or anxious partners, were found to behave more negatively toward their partners during a stress inducing task (Campbell et al., 2001). Similarly, partner effects were found in a study of dating couples indicating that individuals with an anxious partner reported negative interactions and low intimacy in their relationships (Bradford, Feeney, & Campbell, 2002). Such findings indicate that the behavior of one individual may elicit certain behaviors from his or her partner. The existence of partner effects is theoretically consistent with attachment theory. Bowlby’s (1973) position that repeated interactions with significant others create a lens, or working model, through which individuals view the self and others suggests the repeated actions of one member of the dyad may serve to confirm her partner’s working model and in turn influence her behavior (Furman & Simon, 2006).

Current study and hypotheses

The current study compares adolescents, aged 15–18 years, on attachment-related outcomes in their relationships with romantic partners.

H1: Attachment anxiety and avoidance will be positively related to risky behavior among adolescents in romantic relationships.

Although avoidant youth have demonstrated both elevated and inhibited risk behavior compared to secure and anxious youth (Kassel et al., 2007), studies suggest avoidant youth may participate in risk behavior in order to cope with distress in intimate situations (Cooper et al., 1998). All of the adolescents in this study were involved in an intimate relationship; therefore, both attachment anxiety and avoidance were expected to be positively related to risk behavior.

H2: Individuals with partners high in attachment avoidance or anxiety will engage in more risk behavior than individuals with partners low in avoidance or anxiety.

Studies on romantic relationships have often failed to obtain data from both members of a relationship dyad (Campbell et al., 2001; Furman & Simon, 2006). Therefore, limited research is available to assess how one partner’s attachment characteristics may influence the behavior of his or her partner. Given previous research suggesting that the characteristics of one member of the dyad may elicit certain behaviors from his or partner (Bradford et al., 2002), the characteristic behavior of certain attachment orientations, such as the extreme need for intimacy of anxious individuals (Tracy et al., 2003), was predicted to influence the behaviors of partners. For example, adolescents with highly anxious partners may engage in sexual activity to satisfy their anxious partner’s need for intimacy.

H3: Self-reported romantic attachment avoidance and anxiety will be positively related to observed attachment avoidance and anxiety.

The present study included multiple measurements of attachment from multiple perspectives allowing for an assessment of the reliability of self-report data in attachment research. An observational measure of romantic attachment and self-reported romantic attachment were compared to assess the accuracy of reporting among adolescents, which has been of interest especially among young adolescent samples.

Method

Participants

Risk behavior and attachment relationships with romantic partners were assessed among adolescent couples (n = 21). Adolescents and their romantic partners (N = 42) were recruited using flyers, and newspaper and online advertisements between January 2011 and December 2011. To be eligible for the study, adolescents had to be in the ages between 15 and 18 years, living with at least one parent, and currently involved in a romantic relationship with a partner who was also willing to participate. A romantic relationship was defined as a mutually recognized, ongoing relationship between a male and a female lasting at least one continuous month.

Procedure

Interested participants contacted project staff by phone or e-mail. A brief screening form was administered by staff to each potential participant to determine eligibility. If the adolescent expressed interest, his or her romantic partner was contacted to assess the partner’s eligibility. Once eligibility and interest in both members of the couple was determined, the parents of the couple were contacted within 24 hours in order to obtain written parental consent for those participants under the age of 18 years. Of the 72 participants approached, 48% (n = 35) were eligible and 29% (n = 21) were enrolled in the study. Primary reasons for ineligibility included not having a current partner, or having a partner outside of the specified age range. Romantic partners’ lack of time or interest in participating and inability or refusal to contact parents for parental consent were the main reasons of the eligible youth who declined to participate.

An appointment was then set to meet with the couple to obtain written assent/consent and to complete the assessment battery. At the participants’ preference, the assessment interview took place in either a private office on campus, a local library, or in the participants’ homes. In order to maintain confidentiality between partners and reduce response bias, participants completed questionnaires in separate rooms. The full assessment lasted approximately 1 hour, and each participant was compensated $20 cash for his/her time. All procedures were approved by the University’s Institutional Review Board.

Measures

Each adolescent completed a Demographic Questionnaire that assessed adolescent as well as parental characteristics including age, gender, ethnicity, and education status in order to identify any differences in attachment or risk behavior outcomes based on demographic data.

Risk behavior

A combination of one section of the Risk Assessment Battery (Metzger et al., 1991) and portions of the Health Risk Questionnaire (Johnson, Aschkenasy, Herbers, & Gilenwater, 1996; Kann, Nelson, Jones, & Kolbe, 1989) were used to assess sexual activity among the adolescents. The 12-item scale includes questions regarding number of sexual partners, condom use, engaging in sex while under the influence of substances, and sexually transmitted infections. Responses were rated on a scale from 0 (Never) to 3 (Three or more times) with items reverse scored for condom use. All items were summed to create an overall sexual risk score. The Risk Assessment Battery has been used successfully with adolescent populations in previous research (Bassols, Santos, Rhode, & Pechansky, 2007). Good concurrent validity and test–retest reliability have also been demonstrated (Metzger et al., 1991).

A portion of the Teen Addiction Severity Index (T-ASI; Kaminer, Bukstein, & Tarter, 1991) was used to assess substance use. The structured interview addresses seven domains; however, only the substance use domain was administered to assess the frequency and amount of lifetime drug and alcohol use as well as use in the past 30 days. The measure has demonstrated good convergent validity when compared with similar measures of problem behavior as well as good inter-rater reliability with r = .85 for the substance use domain (Kaminer et al., 1991).

Due to the large number of nonsubstance using adolescents in the sample, a risk behavior composite score was created by combining the risk scores from the sexual risk questionnaire and the T-ASI. The sum from the sexual risk questionnaire was combined with a substance use risk score which was created by assigning 1 point for every type of substance used, 1 point for use that began before the age of 15 years, and 1 point for use that occurred more than three times per month. Previous studies with adolescents indicate that engaging in substance use before the age of 15 years (Anthony & Petronis, 1995) and more than three times per month (Tucker, Ellickson, Collins, & Klein, 2006) is more characteristic of problematic substance use rather than general experimentation. The risk behavior composite score ranged from 0 to 13 (M = 5.3; SD = 3.4) with higher scores indicating more risk behavior than lower scores.

Romantic attachment

Self-report

Each adolescent completed the Experiences in Close Relationships-Revised Questionnaire (ECR-R; Fraley, Waller, & Brennan, 2000) to report romantic attachment to his or her partner. The ECR-R consists of 36 items describing behavior in relationships with significant others rated on a 7-point likert scale. The dimensions of avoidance and anxiety have previously shown an excellent internal consistency (Fraley et al., 2000) and good convergent and discriminant validity (Sibley, Fischer, & Liu, 2005). Cronbach’s alphas for the current study were .80 for the anxiety subscale and .84 for the avoidance subscale. The means and SDs for the avoidance and anxiety dimensions were 2.6 (.9) and 2.9 (.9), respectively.

Observational

The Autonomy and Relatedness coding system (Allen, Hauser, Bell, McElhaney, & Tate, 2003) was used to assess a 10 minute video-taped interaction task. Participant interaction was rated according to a 10 category coding scheme evaluating behaviors promoting and undermining autonomy and relatedness (Allen, Porter, McFarland, McElhaney, & Marsh, 2007). Behaviors promoting autonomy (e.g., justifying one’s position with coherent reasons, speaking with a confident tone) and relatedness (e.g., making validating statements, showing empathy) are consistent with concepts relating to low attachment avoidance and anxiety. Likewise, behaviors undermining autonomy and relatedness such as interrupting or ignoring one’s partner or pressuring one’s partner to agree are similar to the pressure for intimacy associated with attachment anxiety and the need for distance characteristic of attachment avoidance. Following procedures outlined in the coding manual, an autonomy, and relatedness composite score was created for each member of the dyad (Allen, Hauser, Bell, Boykin, & Tate, 1994; Scharf & Mayseless, 2008). The score (M = 3.7; SD = 3.4) was computed by subtracting the scores on the undermining autonomy and relatedness scales from scores on the promoting autonomy and relatedness scales. High autonomy and relatedness scores are characteristic of attachment security, while low scores indicate behaviors associated with attachment insecurity.

The topic of discussion for the video-taped session was chosen from the Areas of Change Questionnaire (ACQ; Margolin, Talovic, & Weinstien, 1983). The ACQ consists of 34 questions addressing specific behaviors rated on a 7-point scale that couples would like to increase or decrease in their relationships. Items that couples moderately disagreed on were selected for discussion. By focusing on a specific disagreement, the adolescents’ stress levels were expected to be elevated, thus activating attachment behavior and allowing for a proxy measure of observational romantic attachment (Mangelsdorf & Frosch, 2000). Adolescents most often discussed dissatisfaction with the amount of time their partners spent playing video games or engaging with social media (e.g., Facebook, texting), and jealousy related to their partners talking to members of the opposite sex.

All videos were subsequently transcribed and coded according to the Autonomy and Relatedness coding scheme (Allen et al., 2003) by two coders trained in the system. The first coder coded all videos while the second coder coded 50% of the videos in order to assess inter-rater reliability. Coders were trained over a 3-month-period using materials provided by the system’s authors. Practice continued until inter-rater reliability reached .75. In the current study, inter-rater reliability was good (Intra-class correlation (ICC) = .77).

Data analysis plan

Recent advances in statistical techniques have allowed researchers to analyze dyadic data with complex modeling techniques such as structural equation modeling (SEM) and hierarchical linear modeling (HLM). However, a sample size of 100 dyads or more is required for these advanced methods (Kline, 2005) making the analyses difficult to utilize for researchers using pilot data or small clinical samples. In cases of limited sample size, a pooled regression technique is preferable (Tambling, Johnson, & Johnson, 2011).

In a pooled regression approach, two regression equations are estimated – one on the individual level (i.e., within-dyad effects) and one on the dyad level (i.e., between-dyad effects). The within dyad regression equation consists of difference scores between the partners on the independent and dependent variables. The between dyad regression equation is comprised of average scores on the independent and dependent variables. Next, the path coefficients of the two regression equations are used to compute the actor and partner effects. Actor effects are calculated as the average of the within-dyad and between-dyad coefficients. Partner effects equal the difference between the within-dyad and between-dyad coefficients, divided by two. Pooled standard errors and pooled degrees of freedom must then be estimated in order to calculate the t statistic for assessing the significance level of the actor and partner effects (see Tambling et al., 2011).

In the first set of analyses, attachment anxiety and avoidance for each partner was used to predict one’s own risk behaviors (actor effect) as well as his/her partner’s risk behaviors (partner effect). The association between self-reported and observed attachment behavior was also estimated with the pooled regression approach. Self-reported attachment anxiety was used to predict personal and partner autonomy and relatedness in the first model. The same process was completed for the second model with self-reported attachment avoidance predicting autonomy and relatedness.

Power

A statistical power analysis was conducted for the study using G*Power 3.1.6 Software (Faul, Erdfelder, Buchner, & Lang, 2009). Estimated effect sizes were used in the power calculation. Given a sample of 21 dyads, and an alpha level of .10, the models have approximately 73% power to detect a large effect size, and approximately 42% power to detect a medium effect size.

Results

Sample characteristics

The average age of the adolescents was 17.2 years (SD = .94). Self-reported ethnicity was 54.7% White, 35.7% African American, 4.8% Asian American, and 4.8% other. On average, adolescents reported dating exclusively for over 1 year (M = 15.4 months, SD = 7.7), with relationship lengths ranging from 3 to 30 months. Nearly half of the adolescents (42.8%) reported living with both biological parents, and half of the adolescents’ parents had a postsecondary degree. Adolescents reported using substances for the first time at the age of 14 years, on average. Half of the youth reported having one lifetime sexual partner while 12% reported no sexual intercourse. Demographic characteristics are presented in Table 1.

Table 1.

Demographic characteristics (N = 42).

	M (SD)	Percentage	Range
Male		50.0
Age (years)	17.2 (.9)		15–18
15		7.1
16		14.3
17		31.0
18		47.6
Ethnicity
White		54.7
African American		35.7
Asian American		4.8
Other		4.8
Relationship length (months)	15.4 (7.7)		3–30
GPA	3.1 (.6)		1.9–4.0
Birth parent marital status
Never married		16.7
Married		42.8
Separated		2.4
Divorced		26.2
Widowed		7.1
Unknown^a		4.8
Parent highest degree
<HS diploma		2.4
HS diploma		40.5
2-year degree		23.8
Bachelors degree		26.2
Masters degree		7.1

GPA: grade point average; HS: high school.

^aTwo adolescents were adopted and did not know their birth parents.

Males and females were compared on study variables using paired t-tests. Significant gender differences were found for risk behaviors. Specifically, males (M = 6.43, SD = 3.53) reported higher scores on the risk behavior composite score than females (M = 4.00, SD = 2.79, p = .003). No significant differences in attachment avoidance, attachment anxiety, or autonomy and relatedness were identified between males and females.

Primary analysis

Prior to analysis, all study variables were assessed for adherence to the assumptions of regression. The independent variables (i.e., attachment avoidance and anxiety) were mean centered in order to give zero a meaningful value and to aid in the interpretation of the results (Aiken & West, 1991). Due to the exploratory nature of the study and the limited sample size, a more liberal alpha level, .10, was selected in order to identify relationships that may be missed at the conventional .05 level (King, Hartke, & Denby, 2007).

H1: Attachment anxiety and avoidance will be positively related to risky behavior among adolescents in romantic relationships.

In the first model, attachment anxiety was used to predict individual risk behavior. The actor effect for attachment anxiety was .96. The actor effect was significantly different from zero with a t-value of t(30) = 1.67, p = .10. The effect was in the hypothesized direction and can be interpreted as for every 1 point above the mean an individual scored in attachment anxiety, his or her risk behavior increased by .96. The analysis was repeated using attachment avoidance to predict the individual’s risk behavior. No significant actor effects were found indicating that self-reported attachment avoidance was not related to risk behavior.

H2: Partners high in attachment avoidance or anxiety will engage in more risk behavior than individuals with partners low in avoidance or anxiety.

Next, partner effects were estimated. The partner effect for attachment anxiety was .76. The effect was not significant (t(30) = 1.32, p = .19) suggesting that one’s own attachment anxiety was related to his or her own risk behavior, but not his or her partner’s risk behavior. The analysis was repeated using attachment avoidance as the independent variable; however, no significant partner effects were found. Results indicated that the individual’s own attachment avoidance was not related to his or her partner’s participation in risk behavior.

H3: Self-reported romantic attachment avoidance and anxiety will be positively related to observed attachment avoidance and anxiety.

Two models, using the autonomy and relatedness composite score as the dependent variable, were analyzed. The first model assessed the relationship between attachment anxiety and autonomy and relatedness, and the second model used attachment avoidance as the independent variable. The first model found no significant actor or partner effects for the relationship between anxiety and autonomy and relatedness. In the second model, using avoidance as the independent variable, actor effects approached significance (t(24) = −1.63, p = .11). The trend suggests that individuals with higher self-reported attachment avoidance may be less likely to engage in behaviors that promote autonomy and relatedness (i.e., secure behaviors) in their relationships than individuals who score lower in self-reported attachment avoidance. In addition, significant partner effects were found (t(24) = −2.57, p = .02) suggesting an individual’s autonomy and relatedness score was lower when his/her partner was more avoidant compared to less avoidant (see Table 2 for the results of all regression analyses).

Table 2.

Results of pooled regression by dependent variable.

	Anxiety			Avoidance
	b	t	p	b	t	p
Risk behavior
Actor	.96	1.67	.10	.26	.41	.68
Partner	.76	1.32	.19	.21	.33	.74
AR composite
Actor	.25	.39	.69	−1.00	−1.63	.11
Partner	−.03	−.05	.96	−1.53	−2.57	.02

Note. AR composite = autonomy and relatedness composite score.

Discussion

Although increasing, research focused on adolescent attachment in early romantic relationships is limited. Understanding the behaviors of adolescents in these early relationships is important as youth often begin experimenting with substances and sexual activity within the context of a romantic relationship. This is one of the few studies assessing romantic attachment among both members of adolescent couples while utilizing both self-report and observational measures. However, all results must be viewed as preliminary given the small sample size and exploratory nature of the study.

The first hypothesis predicted a positive relationship between insecure attachment, either avoidant or anxious, and risk behavior. The hypothesis was partially supported with adolescents who scored high in attachment anxiety participating in more frequent substance use and sexual activity compared to those who scored low in anxiety. The result is consistent with previous research suggesting a link between attachment anxiety and frequent involvement in risk behaviors among youth (Cooper et al., 1998). It is possible that anxious youth engaged in more sexual activity in order to increase their sense of intimacy in the relationship, or used substances to cope with fears of rejection (Golder et al., 2005). Yet, further research is needed to identify the causal mechanisms responsible for the relationship.

Interestingly, a relationship between attachment avoidance and risk behavior was not found. Individuals who reported high avoidance on attachment measures have previously reported either heightened or reduced levels of risk behavior due to their discomfort with intimacy. In some cases, avoidant youth engaged in frequent risk behavior as a way to withdraw emotionally from peers (Tracy et al., 2003), while others limited their exposure to risk behavior by physically withdrawing from social situations (Cooper et al., 1998). Possibly, differences in the frequency of risk behaviors reported among avoidant youth resulted in the absence of a linear relationship in the data.

The second hypothesis predicted a relationship between one’s own attachment and his or her partner’s risk behavior (i.e., a partner effect). Neither attachment anxiety nor avoidance was related to one’s partner’s risk behavior in the current sample. The limited sample size may have influenced the power to detect partner effects. Partner effects are often smaller than actor effects as actor effects are more direct than partner effects (Furman & Simon, 2006; Kenny & Cook, 1999). Therefore, future studies with larger samples may be more likely to reveal a relationship between one’s attachment and his or her partner’s participation in risk behavior.

The third hypothesis tested the relationship between self-reported romantic attachment and observed attachment. A trend was found for the relationship between self-reported attachment avoidance and observed attachment insecurity (i.e., low autonomy and relatedness). Specifically, higher self-reported avoidance was associated with observed attachment insecurity suggesting that youth who self-reported attachment avoidance also displayed insecure behaviors during the observation. Previous research has indicated a tendency for avoidant individuals to idealize themselves and their relationships (Main, 1996), which may bias the accuracy of their self-report responses. However, the finding of a relationship between self-reported avoidance and observed insecure behaviors offers some support for the use of self-reported attachment avoidance with adolescent samples.

A significant partner effect emerged indicating that individuals with highly avoidant partners displayed more insecure behaviors during the interaction task. Researchers posit that the attachment orientations of each member of a couple may interact and influence each other (Feeney, 2003). For example, over time, a partner with a secure attachment may reduce his/her anxious partner’s fear of abandonment by consistently responding to his/her needs. In the current study, the desire for distance among avoidant individuals possibly created frustration in their partners leading them to respond with insecure behaviors; however, causal and temporal relationships cannot be concluded from this study.

Contrary to predictions, attachment anxiety was not related to autonomy and relatedness in the observation task. The finding may indicate difficulties among youth in self-reporting attachment anxiety, poor validity of the anxiety dimension of the self-report measure among adolescents, or possibly an insensitivity of the observational measure to assess behaviors consistent with attachment anxiety. Given that this was the first study to compare self-reported romantic attachment among adolescents to interactions coded using the Autonomy and Relatedness coding system, measurement error is considered a likely explanation. Perhaps behaviors characteristic of attachment avoidance, such as ignoring one’s partner’s arguments, are more easily identified according to the guidelines of the coding system compared to anxious behaviors such as hostility. Indeed, social desirability may have been a factor in the amount of overt hostility portrayed during the interaction. A video camera was visible in the room as opposed to behind a mirror or other hidden location. Behaviors theoretically associated with anxiety in the coding system are more overtly negative than those associated with avoidance, and the physical presence of the camera may have reduced the likelihood of couples to react with anger or hostility. The couples in this study may be particularly susceptible to social desirability as the period of adolescence is associated with the developmental concept of the “imaginary audience” in which youth feel they are constantly being evaluated (Elkind, 1967).

Limitations

The present study builds on previous research on risk behavior and adolescent attachment while filling notable gaps in the literature; however, some limitations must be acknowledged. First, the limited sample size posed several challenges. The statistical power to detect differences was low due to the number of participants; however, the current study was intended as an exploratory study to provide a unique dataset to the field and serve as a foundation for future studies. Additionally, sophisticated statistical tools such as SEM or HLM commonly recommended for dyadic data analyses were not appropriate given the size of the sample. Yet, given the barriers related to gathering data from both members of young adolescent couples (e.g., obtaining parental consent, short duration of relationships), the current study provides an initial glimpse into the attachment behaviors of adolescents younger than college age; a population often under-represented in adolescent attachment research (Cooper et al., 1998). Another limitation due to sample size involves the lack of covariates in the analysis. Future research is needed to identify whether the observed relationships are due to attachment or other explanatory variables.

Second, the risk behavior composite score has not been previously validated with other samples and was likely driven by sexual risk behavior rather than substance use due to the unequal number of items comprising the composite score. The researchers focused on face validity for this exploratory study and created the items based on relevant research findings. Additionally, the adolescents in the study did not constitute a representative sample. Participants contacted project staff expressing interest in the study; therefore, the nature of the recruitment process may have predisposed the sample to include mostly high-functioning couples. Indeed, most of the youth had been in their relationship for over 1 year, and appeared to have more agreeable relationships with parents and romantic partners. The sample also included only heterosexual couples. Therefore, the results may be most generalizable to long-term, heterosexual couples.

The sample exhibited minimal substance use possibly leading to a floor effect, which may have limited the ability to identify relationships between attachment and risk behavior. And finally, given that participants were assessed at only one time point, no causal inferences can be made regarding whether attachment leads to risk behavior.

Implications

Clinical

The possible link between attachment anxiety and risk behavior suggests prevention programs targeted at reducing risk behaviors may benefit from assessing attachment among youth in order to identify those who may be at increased risk of engaging in dangerous behaviors. Also, intervention programs targeting substance abuse may improve if issues associated with feelings of rejection and abandonment are addressed among those reporting high anxiety.

Furthermore, by treating only the individual presenting symptoms, clinicians may be missing an intervention opportunity. Individuals with avoidant partners were more likely than those with less avoidant partners to respond with behaviors that decreased autonomy and relatedness during the interaction task. According to attachment theory, avoidant individuals’ discomfort with intimacy may lead them to expect the negative responses they receive from partners, which serves to confirm their attachment insecurity (Campbell et al., 2001). Clinicians who suspect their client’s expectation of negativity during interactions with partners may be able to intervene at the appropriate times and stop detrimental relational processes from continuing.

Methods for assessing adolescent attachment

The present study provides preliminary evidence of adolescents’ ability to self-report their attachment behaviors, at least regarding attachment avoidance. Preliminary support for the use of self-report attachment questionnaires with adolescents offers some validation for the conclusions reached in previous research utilizing the ECR-R with youth. However, some caution should be taken when interpreting these results as half (47.6%) of the youth in this study was at the top of the specified age range (15–18 years). Accurate interpretation of, and response to, questions on the attachment survey may differ for younger adolescents. Still, self-report attachment measures may offer researchers at least an initial estimate of adolescent attachment in a brief, inexpensive format.

Although a relationship between self-reported attachment anxiety and greater observed insecurity in the observation task was not found, measurement error may have influenced the result. Attachment anxiety was expected to be related to overtly hostile statements and actions; however, adolescents may have been less likely to engage in negative behavior due to discomfort with being filmed. Future research assessing observed attachment using the Autonomy and Relatedness coding system may benefit from placing the camera more discretely in the room, or using a warm-up task to provide the youth with more time to acclimate to the idea of being recorded (Gross, 1991).

Conclusions

The results of this exploratory study provide justification for the continued study of adolescent romantic attachment within a larger, more representative sample. Romantic attachment anxiety was related to risk behavior in the actor indicating a potential benefit of addressing attachment in risk prevention and intervention programs. Partial support for the validity of a popular self-report attachment measure was found providing preliminary justification for the use of self-report surveys with adolescents; greatly reducing the expense and time associated with interviews. However, future research is needed to fully confirm the validity of adolescent self-reported attachment. Because individuals often begin experimenting with risk behavior during adolescence, prevention and intervention programs targeting them are especially salient.

Footnotes

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

References

Aiken

L. S.

West

S. G.

(1991). Multiple regression: Testing and interpreting interactions. Newbury Park, CA: Sage Publications.

Allen

J. P.

Hauser

S. T.

Bell

K. L.

Boykin

K. A.

Tate

D. C.

(1994). Autonomy and relatedness coding system manual, version 2.01. Unpublished manual.

Allen

J. P.

Hauser

S. T.

Bell

K. L.

McElhaney

K. B.

Tate

D. C.

(2003). Autonomy and Relatedness Coding System Manual Version 2.14. University of Virginia.

Allen

J. P.

Porter

McFarland

F. C.

McElhaney

K. B.

Marsh

(2007). The relation of attachment security to adolescents' paternal and peer relationships, depression, and externalizing behavior. Child Development, 78, 1222–1239. doi:10.111/j.1467-8624.2007.01062.x

Anthony

J. C.

Petronis

K. R.

(1995). Early-onset drug use and risk of later drug problems. Drug and Alcohol Dependence, 40, 9−15. doi:10.1016/j.addbeh.2008.10.021

Bassols

A. M.

Santos

R. A.

Rhode

L. A.

Pechansky

(2007). Exposure to HIV in Brazilian adolescents: The impact of psychiatric symptomatology. European Child and Adolescent Psychiatry, 16, 236–242. doi:10.1007/s00787-006-0595-7

Bowlby

(1969). Attachment and loss, Vol. 1: Attachment. New York, NY: Basic Books.

Bowlby

(1973). Attachment and loss, Vol. 2: Separation. New York, NY: Basic Books.

Bradford

S. A.

Feeney

J. A.

Campbell

(2002). Links between attachment orientations and dispositional and diary-based measures of disclosure in dating couples: A study of actor and partner effects. Personal Relationships, 9, 491–506. doi:10.1111/1475- 6811.00031

10.

Campbell

Simpson

J. A.

Kashy

D. A.

Rholes

W. S.

(2001). Attachment orientations, dependence, and behavior in a stressful situation: An application of the actor-partner interdependence model. Journal of Social and Personal Relationships, 18, 821–843. doi:10.1177/0265407501186005

11.

Centers for Disease Control. (2012). Sexual risk behavior: HIV, STD, and teen pregnancy prevention. Retrieved March 8, 2013: http://www.cdc.gov/healthyyouth/sexualbehaviors/index.htm

12.

Centers for Disease Control and Prevention. (2011). Sexual identity, sex of sexual contacts, and health-risk behaviors among students in grades 9–12: Youth risk behavior surveillance, selected sites, United States, 2001–2009. MMWR Early Release, 60, 29–33.

13.

Collins

W. A.

Welsh

D. P.

Furman

(2009). Adolescent romantic relationships. Annual Review of Psychology, 60, 631–652. doi:10.1146/annurev.psych.60.110707.163459

14.

Cooper

M. L.

Shaver

P. R.

Collins

N. L.

(1998). Attachment styles, emotion regulation, and adjustment in adolescence. Journal of Personality and Social Psychology, 74, 1380–1397. doi:10.1037//0022-3514.74.5.1380

15.

Eaves

C. S.

(2004). Heroin use among female adolescents: The role of partner influence in path of initiation and route of administration. American Journal of Drug and Alcohol Abuse, 30, 21–38. doi:10.1081/ADA-120029864

16.

Elkind

(1967). Egocentrism in adolescence. Child Development, 38, 1025–1034.

17.

Faul, F., Erdfelder, E., Buchner, A., & Lang, A. G. (2009). Statistical power analyses using G*Power 3.1: Tests for correlation and regression analyses. Behavior Research Methods, 41, 1149–1160.

18.

Feeney

J. A.

(2003). The systemic nature of couple relationships: An attachment perspective. In Erdman

Caffery

(Eds.), Attachment and family systems: Conceptual, empirical, and therapeutic relatedness (pp. 139–163). New York, NY: Brunner/Mazel.

19.

Feldstein

S. W.

Miller

W. R.

(2006). Substance abuse and risk-taking among adolescents. Journal of Mental Health, 15, 1–11.

20.

Fraley

R. C.

Waller

N. G.

Brennan

K. A.

(2000). An item-response theory analysis of self- report measures of adult attachment. Journal of Personality and Social Psychology, 78, 350–365. doi:10.1037//0022-3514.78.2.350

21.

Furman

Simon

V. A.

(2006). Actor and partner effects of adolescents’ romantic working models and styles on interactions with romantic partners. Child Development, 77, 588–604. doi:10.1111/j.1467-8624.2006.00892.x

22.

Gentzler

A. L.

Kerns

K. A.

(2004). Association between insecure attachment and sexual experiences. Personal Relationships, 11, 249–265. doi:10.1111/j.1475-6811.2004.00081.x

23.

Golder

Gillmore

M. R.

Spieker

Morrison

(2005). Substance use, related problems, behaviors and adult attachment in a sample of high risk older adolescent women. Journal of Child and Family Studies, 14, 181–193. doi:10.1007/s10826-005-5046-6

24.

Gross

(1991). Issues related to validity of videotaped observational data. Western Journal of Nursing Research, 12, 658–663. doi:10.1177/019394599101300511

25.

Hazan

Shaver

P. R.

(1987). Romantic love conceptualized as an attachment process. Journal of Personality and Social Psychology, 52, 511–524. doi:10.1037%2F%2F0022-3514.52.3.511

26.

Johnson

Aschkenasy

Herbers

Gillenwater

(1996). Self-reported risk factors for AIDS among homeless youth. AIDS Education and Prevention, 8, 308–322.

27.

Johnston

L. D.

O’Malley

P. M.

Bachman

J. G.

Schulenberg

J. E.

(2008). Monitoring the Future national results on adolescent drug use: Overview of key findings, 2007 (NIH Publication No. 08-6418). Bethesda, MD: National Institute on Drug Abuse.

28.

Kaminer

Bukstein

O. G.

Tarter

R. E.

(1991). The teen-addiction severity index: Rationale and reliability. International Journal of Addiction, 26, 219–226.

29.

Kann

Nelson

Jones

Kolbe

(1989). Establishing a system of complementary school-based surveys to annually assess HIV-related knowledge, beliefs, and behaviors among adolescents. Journal of School Health, 59, 55–58. doi:10.1111/j.1746- 1561.1989.tb05392.x

30.

Karney

B. R.

Beckett

M. K.

Collins

R. L.

Shaw

R. N.

(2007). Adolescent romantic relationships as precursors of healthy adult marriages: A review of theory, research, and programs. (TR433-ACF). Santa Monica, CA: RAND Corporation.

31.

Kassel

J. D.

Wardle

Roberts

J. E.

(2007). Adult attachment security and college student substance use. Addictive Behaviors, 32, 1164–1176. doi:10.1016/j.addbeh.2006.08.005

32.

Kenny

D. A.

Cook

W. L.

(1999). Partner effects in relationship research: Conceptual issues, analytic difficulties, and illustrations. Personal Relationships, 6, 433–448. doi:10.1111/j.1475-6811.1999.tb00202.x

33.

Kenny

D. A.

Kashy

D. A.

Cook

W. L.

(2006). Dyadic data analysis. New York, NY: Guilford Press.

34.

Kershaw

T. S.

Milan

Westdahl

Lewis

Rising

S. S.

Fletcher

Ickovics

(2007). Avoidance, anxiety, and sex: The influence of romantic attachment on HIV-risk among pregnant women. AIDS Behavior, 11, 299–311. doi:10.1007/s10461-006-9153-z

35.

King

R. B.

Hartke

R. J.

Denby

(2007). Problem-solving early intervention: A pilot study of stroke caregivers. Rehabilitation Nursing, 32, 68–76. doi:10.1002/j.2048-940.2007.tb00154.x

36.

Kline

R. B.

(2005). Principles and practice of structural equation modeling (2nd ed.). New York, NY: Guilford.

37.

Main

(1996). Introduction to the special section on attachment and psychopathology: 2. Overview of the field of attachment. Journal of Consulting and Clinical Psychology, 64, 237–243. doi:10.1037/0022-006X.64.2.237

38.

Mangelsdorf

S. C.

Frosch

C. A.

(2000). Attachment and temperament: One construct or two? In Reese

H. W.

(Ed.), Advances in child development and behavior (pp. 182–214). San Diego, CA: Academic Press.

39.

Margolin

Talovic

Weinstein

C. D.

(1983). Areas of change questionnaire: A practical approach to marital assessment. Journal of Consulting and Clinical Psychology, 51, 920–931. doi:10.1037/0022-006X.51.6.920

40.

Metzger

D. S.

Woody

G. E.

Druley

O'Brien

C. P.

McLellan

A. T.

Williams

J.,

… O’Brien

, C. P.

(1991). Psychiatric symptoms, high risk behaviors and HIV positivity among methadone patients. In Harris

(Ed.), Problems of Drug Dependence (pp. 490–491). NIDA Research Monograph No. 105. Rockville, MD: NIDA.

41.

Office of Applied Studies. (2007). Results from the 2006 national survey on drug use and health: National findings (DHHS Publication No. SMA 07-4293, NSDUH Series H-32). Rockville, MD: Substance Abuse and Mental Health Services Administration.

42.

Reese-Weber

Marchand

J. F.

(2002). Family and individual predictors of late adolescents’ romantic relationships. Journal of Youth and Adolescence, 31, 197–207.

43.

Schachner

D. A.

Shaver

P. R.

(2004). Attachment dimensions and sexual motives. Personal Relationships, 11, 179–195. doi:10.1111/j.1475-6811.2004.00077.x

44.

Scharf

Mayseless

(2008). Late adolescent girls' relationship with parents and romantic partners: The distinct role of mothers and fathers. Journal of Adolescence, 31, 837–855. doi:10.1016/j.adolescence.2008.06.012

45.

Sibley

C. G.

Fischer

Liu

J. H.

(2005). Reliability and validity of the revised experiences in close relationships (ECR-R) self-report measure of adult romantic attachment. Personality and Social Psychology Bulletin, 31, 1524–1536. doi:10.1177/0146167205276865

46.

Strachman

Impett

E. A.

, (2009). Attachment orientations and daily condom use in dating relationships. The Journal of Sex Research, 46, 1–11. doi:10.1080/00224490802691801

47.

Tambling

R. B.

Johnson

S. K.

Johnson

L. N.

(2011). Analyzing dyadic data from small samples: A pooled regression actor-partner interdependence model approach. Counseling Outcome Research and Evaluation, 2, 101–114. doi:10.1177/2150137811422901

48.

Tracy

J. L.

Shaver

P. R.

Albino

A. W.

Cooper

(2003). Attachment styles and adolescent sexuality. In Florsheim

(Ed.), Adolescent romantic relations and sexual behavior: Theory, research, and practical implications (pp. 137–159). Mahwah, NJ: Lawrence Erlbaum Associates.

49.

Tucker

J. S.

Ellickson

P. L.

Collins

R. L.

Klein

D. J.

(2006). Are drug experimenters better adjusted than abstainers and users? A longitudinal study of adolescent marijuana use. Journal of Adolescent Health, 39, 488–494. doi:10.1016/j.jadohealth.2006.03.012

50.

Weinfield

N. S.

Whaley

G. L. B.

Egeland

(2004). Continuity, discontinuity, and coherence in attachment from infancy to late adolescence: Sequelae of organization and disorganization. Attachment and Human Development, 6, 73–97. doi:10.1080/14616730310001659566

51.

Zimmer-Gembeck

M. J.

Siebenbruner

Collins

W. A.

(2004). A prospective study of intraindvidual and peer influences on adolescents’ heterosexual romantic and sexual behavior. Archive of Sexual Behavior, 33, 381–394. doi:10.1023/B:ASEB.0000028891.16654.2c