Romantic Relationship Trajectories Among Young African American Men: The Influence of Adverse Life Contexts

Abstract

The development of supportive and committed romantic relationships during emerging adulthood forecasts relationship quality later in adulthood. Many emerging adult African American men are exposed to challenging socioeconomic environments that are known to undermine romantic relationships. Studies of African American men’s romantic relationship quality and its antecedents during this developmental period are scarce. The present study investigates longitudinal trajectories of romantic relationship quality among African American emerging adult men and then tests a model linking them to adverse childhood experiences, socioeconomic instability, community disadvantage, and defensive relational schemas. The analytic sample included 331 African American men who reported having a romantic partner, observed during three waves of data collection from ages 19 to age 26. Using men’s reports of romantic relationship support, conflict, and dyadic trust, parallel growth mixture modeling was conducted to identify romantic relationship trajectory profiles. We identified three romantic relationship trajectory profiles: Normative, Uncertain, and Conflictual. Structural equation analyses revealed that adverse childhood experiences were associated positively with contemporaneous contextual risk factors (i.e., socioeconomic instability and community disadvantage), which in turn, were associated significantly with membership in Uncertain and Conflictual trajectories through defensive relational schemas. The present study reveals heterogeneous romantic relationships among African American emerging adult men. Findings support the conjoint influences of early adversity and contemporaneous stressors as robust antecedents of African American men’s romantic relationship behaviors over time.

Developmental theorists identify the formation of nurturing and committed bonds with a romantic partner as a critical task of emerging adulthood (Arnett, 2007); one with prognostic significance for marriage and marriage-like bonds in adulthood (Collins, Welsh, & Furman, 2009; Fincham & Cui, 2010). Romantic bonds in emerging adulthood can have profound influences on individual well-being (Simon & Barrett, 2010; Umberson, Crosnoe, & Reczek, 2010). For example, supportive, nurturing romantic relationships have direct links to reduced risk behavior, enhanced mental health, and vocational development (Barr, Culatta, & Simons, 2013). Supportive romantic relationships also can attenuate the negative influence of adverse childhood experiences and contextual risk factors (e.g., community disadvantage, socioeconomic instability) on young men’s development (Cho & Kogan, 2016).

Considerable empirical attention has been paid to couple relationships among African American women, especially young mothers (Edwards et al., 2012). Rarely, however, do studies focus on young African American men or follow them over time, and extant research with African American men has often focused narrowly on their sexual risk behavior; essentially neglecting relationship quality (Bowleg et al., 2017). Accumulating evidence suggests that disproportionate exposure to socioeconomic stressors takes a toll on African American men’s establishment and maintenance of satisfying romantic relationships (Kurdek, 2008; Raley, Sweeney, & Wondra, 2015). Such stressors undermine men’s likelihood of engaging in close relationships and increase negative affect in established relationships (Cutrona et al., 2003; Kurdek, 2008). Growing up in and navigating challenging environments during the transition to adulthood can take a toll on romantic relationship development and may inform low rates of marriage observed in this population (Barr & Simons, 2012; Dixon, 2009). Despite evidence regarding the importance of romantic relationships and the challenges low SES men experience, prospective investigations of low SES African American men’s romantic relationships and their antecedents are scarce. The present study was designed to address these gaps. We focus specifically on rural African American men who are transitioning to adulthood in communities with high levels of poverty and limited educational and vocational opportunities.

Relationship Quality Trajectories among Emerging Adult African American Men

The first aim of this study was to provide a better understanding of how African American men’s romantic relationship quality changes over time during emerging adulthood. We used a person-centered approach to characterize men’s reported relationship quality over time. Multiple dimensions of romantic relationship processes have been linked to relationship outcomes and individual well-being (Bradbury, Fincham, & Beach, 2000). Informed by existing research (Bryant & Wickrama, 2005; Kelly & Floyd, 2006), we focused on three aspects of relationship functioning that have been shown to be affected by the contextual stressors that African American men experience: relationship conflict, support, and dyadic trust. Relationship conflict and support are key constructs in stress spillover models linking contextual risk factors to African American couples’ functioning (Conger et al., 2002). Under conditions of economic distress and community disadvantage, members of a couple are vulnerable to psychological distress including depressive symptoms, anger, and low frustration tolerance (Pearlin, Schieman, Fazio, & Meersman, 2005). Negative moods induced by contextual risk factors “spill over” to romantic relationships, undermining relationship satisfaction and quality while increasing conflict and aggressive behaviors (Cunradi, Caetano, Clark, & Schafer, 2000).

Stressful and challenging environments also can undermine the routine provision of instrumental and emotional support that contributes to relationship stability and quality (Burton & Tucker, 2009; Conger, Conger, & Martin, 2010). Individuals exposed to stressful life contexts may believe that the economic and emotional returns of engaging in a supportive relationship are minimal (Amato, 2011), which can discourage them from investing time and effort toward their partners and relationships. Previous research on African American couples from low resource communities revealed diminished levels of emotional support compared to individuals from high resource communities (Barden, Barry, Khalifan, & Bates, 2016).

Dyadic trust comprises a distinct relationship dynamic that includes the belief that one’s partner is reliable and caring (Rempel, Holmes, & Zanna, 1985). Trust deepens intimacy and supports long-term commitments and monogamy (Larzelere & Huston, 1980). Studies indicate that for some African Americans, low rates of marriage in the community as a whole (Carlson, McLanahan, & England, 2004) and experiences of harsh parenting (Simons et al., 2012) undermine trust within romantic relationships. Dyadic trust is affected further by rates of and opportunities for concurrent sexual partnerships (Senn et al., 2011). As African American communities have high rates of incarceration and unemployment among men (Kang-Brown & Subramanian, 2017), the number of available “date-able” African American men is reduced (Dixon, 2009), increasing men’s opportunities for concurrent sexual partnerships and undermining partners’ expectations for fidelity and trust in relationships (Nunn et al., 2011).

In the present study, we investigated how relationship conflict, support, and trust characterized men’s perceptions of their romantic relationships over three observations across a three-year period. Men were assigned to relationship trajectory profiles based on the level and change across each variable. The pattern among the variables provides a uniquely informative portrait that can be linked to developmental or health outcomes and used to foster theory and hypothesis development in future studies. Given the lack of similar research with samples of African American men, however, we proffered no a priori hypotheses regarding the specific profiles that would emerge.

Contextual Influences, Defensive Relational Schemas, and Romantic Relationships

Per our second aim, we expected that the romantic relationship quality trajectories which emerged in Aim 1 would be linked to risk factors in men’s rearing environments and in their current context (Figure 1). In the rearing environment, we investigated the role of adverse childhood experiences. Two risk factors were examined in men’s contemporaneous context: (a) socioeconomic instability and (b) community disadvantage. We expected these contextual risk factors to be associated with relationships directly, and indirectly via defensive relational schemas, enduring but modifiable cognitions regarding relationships. Below, we describe theory and evidence for the expectations presented in Figure 1.

Figure 1.

Conceptual model of the antecedents of romantic relationship trajectories

Adverse childhood experiences.

Growing up in low-SES, rural environments places children at heightened risk for exposure to adverse experiences (Health Resources and Services Administration, 2011; Umberson et al., 2014). Adverse experiences include living in destructive home environments, such as those in which children observe intimate partner violence, and experiencing directly neglectful or abusive treatment from caregivers. Adverse childhood experiences represent extreme environmental hazards to psychosocial and cognitive development and predict a wide range of long-term health outcomes and behaviors during young adulthood, including conflict and anger in downstream romantic relationships (Kogan, Yu, & Brown, 2016). We expect that men’s exposure to adverse childhood experiences will predict conflict and low levels of trust and support in emerging adult relationships (Simons, Lin, & Gordon, 1998).

Socioeconomic instability.

Life course perspectives on close relationships emphasize (a) how socioeconomic and other stressors in childhood exhibit continuity into adulthood, and (b) how the accumulation of disadvantage over time takes a toll on the quality of intimate relationships (Umberson et al., 2010). For example, economically or otherwise disadvantaged families may struggle to provide their children with the resources needed to thrive in higher education and to obtain stable careers (Bryant & Conger, 2002). Family economic disadvantage also forecasts adolescents’ and emerging adults’ engagement with school, work, or other institutions that support conventional behavior (Bae & Wickrama, 2019; Hirschi, 1969). Such bonds are linked closely with the quality of close relationships (Giordano, Lonardo, Manning, & Longmore, 2010). We thus expect that childhood adversity will predict socioeconomic instability in emerging adulthood, which in turn will be associated with romantic relationship support, conflict, and trust.

Community disadvantage.

Rural African American men are disproportionately likely to reside in resource-poor neighborhoods characterized by high rates of crime, limited employment opportunities, and weakened community infrastructure (Probst et al., 2002). Childhood adversity forecasts residence in disorganized, low resource communities as young people are less able to marshal the resources to afford more affluent communities or attend college (Williams & Collins, 2001). Disadvantaged communities also provide fewer models for successful relationships and constitute generally stressful environments, both of which negatively affect satisfaction and stability in romantic relationships (Cutrona et al., 2003). In addition, African American young men living in disorganized, resource-poor communities may have limited access to institutional resources, such as counseling and recreational facilities, which can negatively affect the quality of day-to-day interactions (Bryant & Wickrama, 2005). We thus expect childhood adversity to predict community disadvantage, which in turn, will be associated with relationship trajectories.

Relational schemas.

We expect that childhood adversity, socioeconomic instability, and community disadvantage will be associated indirectly with romantic relationship quality trajectories through men’s relational schemas. Relational schemas are enduring beliefs, cognitions, and expectations regarding close relationships (Baldwin, 1995). Developed in response to one’s history of interpersonal interactions with important others, relational schemas help individuals to define situations more efficiently by drawing attention to salient cues in the social environment, goals associated with response options, and consequences associated with particular responses (Crick & Dodge, 1994). Studies indicate that adverse childhood experiences, including inconsistent or chaotic relationships, exert a robust influence on the extent to which men will manifest defensive and mistrustful relational schemas (Kogan, Cho & Oshri, 2016). Underlying perceptions of relationships, however, can be modified in response to environmental circumstances in adolescence and young adulthood (Baldwin, 1995). Unstable, hostile environments, a product of disadvantaged communities and socioeconomic instability, can increase the salience of defensive schemas (Burton & Tucker, 2009). Individuals with defensive schemas are likely to perceive romantic partners as untrustworthy and to exhibit hostile behavior in relationships (Simons et al., 2014), resulting in increased conflict and less intimacy and support.

The Current Study

Despite evidence regarding the importance of romantic relationships in emerging adulthood, prospective studies of rural African American young men’s relationships are scarce. In this study, we first explored relationship quality trajectories considering changes in conflict, support, and trust in men’s romantic relationships over three years using a parallel process growth mixture model. This provided a descriptive portrait of men’s relationship quality over a three-year period. We then examined a model of the antecedents of relationship trajectory. Informed by life course theory, we expected that exposure to early (i.e., adverse childhood experiences) and contemporaneous (i.e., community disadvantage, socioeconomic instability) contextual risk factors would predict relationship quality trajectories directly and indirectly via their influences on defensive relational schemas. This investigation is to our knowledge the first to consider the conjoint contributions of rearing and contemporaneous risk factors on African American men’s romantic relationships during the emerging adult period.

Methods

Participants

Data for this study were from the African American Men’s Project (AMP), a study of relationship development, health risk behavior, and well-being among rural African American men. In 2011, baseline (Wave 1; W1) data were collected from 505 men 19–22 years of age (M = 20.3, SD = 1.08) who were recruited from 11 rural counties in southern Georgia. These counties were selected based on their nonurban designation according to the Census Bureau (e.g., having a population density of less than 100 persons per square mile).

Participants were recruited using respondent-driven sampling (RDS; Heckathorn, 1997). First, community liaisons recruited 45 initial seed participants from the 11 counties. Project staff contacted the referred participants, described the project, determined eligibility, and set up a data-collection visit at the participant’s home or a convenient community site (e.g., a private room in the public library). Then, each of the referred participants provided the names of three men in his personal network who met eligibility criteria. Upon completion of data collection, these participants also provided referral information for three network members. For each network member successfully recruited into the study, the referring participant received $25.

RDS integrates this chain-referral protocol with a mathematical model which enables the analyst to evaluate and, if necessary, attenuate biases commonly associated with network-based samples (Heckathorn, 1997, 2002). These biases include the effects of seeds’ characteristics on the final sample, men’s effectiveness at referring others, and the size of men’s referral networks. Additional information on the sampling network and the effectiveness of the RDS procedures for this study may be found in Kogan et al. (2016). RDS analyses in that study suggested that the procedures minimized biases across a range of demographic and psychosocial characteristics and yielded a sample representative of rural African American men in the target counties.

Data were gathered from participants via audio computer-assisted self-interviews; participants with low literacy were assisted through voice and video enhancements. Each participant received $100 at the conclusion of the data collection visit. The University Institutional Review Board approved all study protocols. Approximately 18.3 months after the baseline survey, a follow-up data collection visit (W2) was conducted in the same manner when men’s age was 21.9 years (SD = 1.27). The third wave of data (W3) was collected 19.7 months after the second wave when men’s age was 23.1 years (SD = 1.26). Of the 505 men who participated at W1, 423 (84%) completed the W2 survey and 409 (81%) completed the W3 survey. Retention in the study was not related to any study predictors (adverse childhood experiences: t = 1.43, p = .16; socioeconomic instability: t = 1.20, p = .23; community disadvantage: t = 1.02, p = .31; and defensive relational schema: t = 1.09, p = .28).

The analytic sample comprised young men who reported they were in a committed romantic or sexual relationships with a main female sexual partner (e.g., a girlfriend or a spouse) at two or more waves of data collection (not necessarily the same female partner), resulting in a final sample size of 331. An analysis was conducted to examine potential differences between our analytic subsample and those who were excluded based on not having a partner. For latent constructs used in the current study (i.e., community disadvantage and defensive relational schema), we used standardized and summed indices to test differences between the two groups. No differences emerged based on study variables (adverse childhood experiences: t = .36, p = .72; socioeconomic instability: t = .96, p = .34; community disadvantage: t = 1.34, p = .18; and defensive relational schema: t = −.88, p = .38).

Measures

Romantic relationship characteristics.

Three romantic relationship variables were assessed from W1 through W3.

Conflict.

Relationship conflict was assessed with a subset of low intensity aggressive and non-aggressive conflict behaviors from the Conflict Tactics Scale-2 (Straus, Hamby, Boney-McCoy, & Sugarman, 1996). The 14-item self-report measure was introduced with instructions to assess the extent to which each item characterizes the ways in which conflict is managed in the relationship. We included items addressing verbal (e.g., insult, yell), psychological (e.g., threaten, damage other’s property), and physical aggression (e.g., hit, slap, or kick) between partners. Items representing nonaggressive, problem-solving forms of conflict resolution (e.g., “I try to stay away from her until I cool down,” “I show respect for my partner’s feelings”) were reverse coded before summing to represent total conflict in the relationships. Responses to each item were scored on a 4-point scale, from 0 (never) to 3 (very often), and the total score was summed with a higher score indicating higher levels of conflict (Cronbach’s α across waves > .80).

Support.

Support provided to one’s romantic partner was assessed with a 3-item subscale of the Network Relationships Inventory (“How often do you take care of her?”, “How often do you protect and look out for her?”, and “How often do you help her with things she can’t do by herself?”; Furman & Buhrmester, 1985). Responses ranged from 0 (never) to 3 (very often), and the total score was summed with a higher score indicating higher levels of support; Cronbach’s α across waves exceeded .81.

Dyadic trust.

Trust was assessed with the 8-item Dyadic Trust Scale (Larzelere & Huston, 1980). Example items include, “There are times when she cannot be trusted,” and “I feel that I can trust her completely.” Responses to each item were scored on a 5-point scale, ranging from 0 (never) to 4 (always). The scale was constructed by summing eight items with a higher score indicating higher levels of trust. Cronbach’s α across waves exceeded .78.

Early and contemporaneous contextual risk factors.

Adverse childhood experiences.

Adverse childhood experiences were assessed using the Adverse Childhood Experiences measure (Felitti et al., 1998) at W1. Participants reported the presence or absence of 10 types of adverse experiences during their first 16 years of life. Adversities included experiencing physical abuse, sexual abuse, or neglect, and witnessing violence to one’s caregiver. Scores ranged from 0 to 10 adversities, with a mean of 2.25 (SD = 2.12). Nationally, 54% of children have not experienced any adverse childhood experiences, and 11% experienced 3 or more adverse childhood experiences (Sacks, Murphye, & Moore, 2014). Among the respondents of the current study, only 23% have not experienced any adverse childhood experiences and 36% experienced 3 or more adverse experiences, suggesting relatively high levels of adverse childhood experiences among respondents of the current study.

Socioeconomic instability.

Socioeconomic instability was assessed using four indicators measured at W1; school enrollment/employment status, unstable living arrangements, low levels of vocational engagement, and high levels of economic distress. Consistent with cumulative models of risk exposure (Evans & Kim, 2010), previous studies suggest that the accumulation of socioeconomic instability factors rather than the influence of any single factor is a robust predictor of developmental outcomes (Rauer et al., 2008). We thus constructed a risk factor index by creating dichotomous variables and summing them. Respondents reported whether they were enrolled in school or employed at a job. Responses were coded as 0 (either enrolled at school or had a job; 46.0%) and 1 (neither enrolled at school nor had a job; 54.0%). Considering the fact that 14 percent of 18- to 24-year-old young adults in the U.S., in general, and 25 percent of African American male, in particular, were neither enrolled in school nor working in 2017 (National Center for Education Statistics, 2019), respondents in the current study showed high levels of unemployment or school non-enrollment.

Unstable living arrangements were assessed with an item regarding changes in living arrangements during the past 6 months. Responses were coded as 0 (1 or 2 residences; 62.8%) or 1 (more than 2 residences; 37.2%). Vocational engagement was assessed using a 10-item scale drawn from a survey that Aseltine and Gore (2005) developed (e.g., “I am a dependable employee,” “I have trouble keeping jobs”). Respondents responded to the items on a scale ranging from 1 (strongly disagree) to 4 (strongly agree); Cronbach’s α was .80. A median split was used to code vocational engagement level (0 = high engagement, 1 = low engagement). Economic distress was assessed with a 5-item scale that captures whether respondents had enough money in the past 3 months for shelter, food, clothing, leisure expenses, and medical expenses (Cho & Kogan, 2016). Responses ranged from 1 (strongly disagree) to 4 (strongly agree); Cronbach’s α was .79. A median split was used to coded economic distress levels (0 = low distress, 1 = high distress). Scores on the four dichotomous indicators were summed to the number of socioeconomic instability risk factors young men experienced, yielding an index that ranged from 0 to 4 (M = 1.69, SD = 1.11).

Community disadvantage.

Community disadvantage was assessed as a latent construct using two scales at W1. The 11-item community disorder measure captures the occurrence of deviant behaviors in the community such as gang fights, robbery or mugging, and rape or other sexual assault (Sampson, Raudenbush, & Earls, 1997). Respondents reported the severity of each problem on a scale ranging from 0 (not a problem) to 3 (a big problem). Cronbach’s alpha for the scale was .90. Low levels of community resources were assessed using a 12-item subscale from the Community Resources and Problems measure (Forehand et al., 2000). Participants responded to the stem, “How good is your community in terms of …,” and example items included “finding full-time jobs” and “finding a place to get help if you have a health problem” The response ranged from 0 (very good) to 4 (very poor), and Cronbach’s alpha was .94. The two scales were significantly associated (r = .59; p < .001).

Defensive relational schemas.

Consistent with previous research (Simons et al., 2012; Kogan et al., 2013), we developed a latent defensive relational schema construct using three scales at W1. The Cynical Views of Relationships scale (Simons, Simons, Lei, & Landor, 2012) is composed of four-items that assess the degree of suspicion that young adults hold toward romantic partners. Example items included “Romantic partners often try to take advantage of me” and “When romantic partners are friendly, they usually want something from you.” The response ranged from 1 (strongly disagree) to 4 (strongly agree). Cronbach’s alpha was .64. The Experiences in Close Relationships scale (Wei, Russell, Mallinckrodt, & Voge, 2007) assesses anxious and avoidant attachment styles. However, we were unable to replicate this factor structure, and our factor analysis (available from the corresponding author) revealed a single “insecure attachment style” scale that combined both avoidance and anxiety items. The resulting six-item scale yielded a Cronbach’s alpha of .74. Example items included, “I try to avoid getting too close to my romantic partners” and “I don’t feel comfortable opening up to romantic partners.” Items were rated on a scale ranging from 1 (strongly disagree) to 4 (strongly agree). The Street Code scale (Stewart & Simons, 2006) is a 7-item scale developed to assess the extent to which an individual believes that aggression is a means of gaining respect in relationships. Example items included, “People tend to respect a person who is tough and aggressive” and “Behaving aggressively is often an effective way of dealing with someone who is taking advantage of you.” Responses ranged from 1 (strongly disagree) to 4 (strongly agree). Cronbach’s alpha was .94.

Demographic covariate.

We controlled target’s age at W1, whether or not the same main partner was reported on, and an average duration of relationships across W1 and W3.

Analytic Approach

Romantic relationship quality profiles were developed using parallel process growth mixture modeling (GMM). This procedure generates latent class profiles using covariation across three separate sets of intercepts (baseline level) and linear slopes (rate of change over time, see Supplemental Figure 1), allowing us to investigate patterns that emerge across these growth parameters. To determine the appropriate number of romantic relationship trajectory classes that best fit the data, we considered several model fit criteria: Bayesian Information Criterion (BIC), sample-size-adjusted BIC (SSA-BIC), and Akaike Information Criterion (AIC). We also utilized the Lo-Mendell-Rubin likelihood ratio test (LMR-LRT), which compares the estimated model to a model with one less class, and entropy, which refers to the average accuracy in assigning individuals to profiles. Participants were assigned to the class in which they had the highest estimated posterior probability. Missing data were managed with full information likelihood estimation (FIML). After the trajectory classes were identified, we tested the model in Figure 1 with multinomial logistic structural equation modeling. We employed the MODEL CONSTRAINT command, which allows for testing indirect effects in models with categorical outcomes. Analyses were conducted in Mplus version 7 (Muthén & Muthén, 2008–2012).

Results

Latent Trajectory Analysis

In a series of GMMs, we observed decreases in AIC and SSA-BIC values as the number of classes increased from one to four, with the exception of the BIC, which increased in the four-class solution relative to the three-class solution (Supplemental Table 1). In addition, decreases in entropy and non-significant LMR-LRT value were observed for the four-class solution. Based on these indices, a 3-class model best represented heterogeneity in romantic relationship quality trajectories. The three trajectory groups are presented in Figure 2.

Figure 2.

Three classes of romantic relationship trajectories among rural African American young men in emerging adulthood

Note. Total analytic sample size is 331. Vertical axis represents estimated romantic relationships characteristics. Horizontal axis represents three survey assessments from wave 1 (W1; mean age = 20.3 years) to wave 3 (W3; mean age = 23.1 years).

The first group (n = 162, 48.9%) characterized nearly half of the participants and was labeled Normative. This trajectory exhibited low levels of relationship conflict at W1 (intercept = 7.24, p < .001) which decreased over time (slope = −.78, p < .001). Men in this trajectory also reported high levels of support (intercept = 6.60, p < .001) and trust (intercept = 23.67, p < .001) compared to other groups; both support and trust increased during the follow-up period (support: slope = .26, p < .05; trust: slope = .96, p < .001).

The second group was labeled Uncertain (n = 127, 38.4%), having a trajectory characterized by a modest level of relationship conflict at W1 (intercept = 9.43, p < .001) which showed no significant change across the follow-up (slope = .07, p = .85). Compared to Normative trajectory class, this trajectory showed lower levels of relationship support (intercept = 5.98, p < .001) and dyadic trust (intercept = 19.86, p < .001), both of which decreased significantly over time (support: slope = −.22, p < .05; trust: slope = −1.25, p < .001).

The final class was labeled the Conflictual trajectory (n = 42, 12.7%) which was characterized by the highest level of relationship conflict at W1 compared to other trajectories (intercept = 11.34, p < .001) which increased over time (slope = 3.48, p < .001). Relationship support was relatively high at W1 (intercept = 6.58, p < .001) but sharply decreased during the follow-up period (slope = −.86, p < .05). Dyadic trust was low at W1 (intercept = 19.78, p < .001) which decreased over time (slope = −1.27, p < .05). This trajectory class exhibited the lowest levels of relationship support and trust at W3 among the three trajectory groups.

To explore potential heterogeneity among respondents who remained in the same relationship at all time points (n = 118) compared to our study sample, we ran an exploratory GMM. As shown in the Supplemental Figure 2, the three trajectory groups that emerged from the analysis are almost identical to the groups which emerged in our main results: Normative (41.8%), Uncertain (30.7%), and Conflictual (27.5%). This result suggests that our primary findings are robust to the influence of relationship status.

Predictors of Relationship Trajectory Class

Table 1 provides descriptive statistics for the respondents in each class. The Conflictual trajectory reported the highest levels of adverse childhood experiences and community disadvantage, and the Uncertain trajectory had the highest levels of socioeconomic instability. The Normative trajectory reported lower levels of contextual risk factors and defensive relational schema than the other two trajectory groups.

Table 1.

Descriptive statistics by romantic relationship trajectory classes at baseline

Variables	Romantic relationship trajectory classes
	Conflictual (n = 42, 12.7%)	Uncertain (n = 127, 38.4%)	Normative (n = 162, 48.9%)	Total (N = 331)
	M (SD) or %	M (SD) or %	M (SD) or %	M (SD) or %
Demographics
Age (years)	20.24 (1.16)	20.22 (1.02)	20.31 (1.10)	20.27 (1.08)
Same main partnership (W1-W3)	38.1%	39.4%	35.2%	37.2%
Average duration of partnership (months)	34.35 (21.91)	31.93 (20.65)	32.56 (23.86)	32.39 (22.60)
Contextual Stressors
Adverse childhood experience	3.14 (2.53)	2.35 (2.12)	1.95 (1.95)	2.25 (2.12)
Socioeconomic instability	1.79 (1.05)	1.89 (1.07)	1.52 (1.13)	1.69 (1.11)
Community disadvantage	.60 (1.41)	.23 (1.54)	−.24 (1.53)	.00 (1.51)
Relationship schemas
Defensive relational schemas	.46 (1.02)	.09 (.85)	−.25 (.96)	.00 (.95)
Romantic relationship characteristics
Conflict (W1)	12.51 (6.18)	9.02 (3.70)	7.39 (4.98)	9.24 (5.50)
Conflict (W2)	16.65 (6.59)	9.34 (4.80)	6.94 (4.75)	9.99 (6.60)
Conflict (W3)	18.19 (4.57)	8.31 (3.24)	5.64 (3.32)	10.28 (6.65)
Support (W1)	6.35 (2.55)	6.30 (2.12)	6.83 (2.16)	6.54 (2.31)
Support (W2)	4.78 (2.28)	5.93 (2.24)	6.93 (2.06)	6.25 (2.31)
Support (W3)	4.43 (2.47)	5.31 (2.23)	7.41 (1.92)	6.22 (2.48)
Dyadic trust (W1)	19.23 (5.36)	18.87 (4.92)	24.63 (4.78)	21.73 (5.70)
Dyadic trust (W2)	17.44 (5.57)	17.99 (5.63)	24.42 (5.46)	21.03 (6.46)
Dyadic trust (W3)	14.96 (3.33)	16.92 (3.24)	27.16 (2.92)	21.02 (6.35)

Note. M = mean; SD = standard deviation.

Prior to testing the indirect effect model (Figure 1), we estimated the direct effects linking adverse childhood experiences to membership in the Conflictual or Uncertain trajectory classes, with Normative as the reference group. The total effect models indicated that adverse childhood experiences were associated significantly with inclusion in the Conflictual trajectory (vs. Normative) (OR [95%] = 1.67 [1.20, 2.34], p < .01). The direct association between adverse childhood experiences and the Uncertain trajectory was not significant (OR [95%] = 1.11 [.86, 1.45], p = .45).

We next tested the pathways linking adverse childhood experiences to relationship trajectories through contemporaneous risk factors and defensive relational schemas (Figure 3). Adverse childhood experiences were associated positively with socioeconomic instability (β = .26, p < .001) and community disadvantage (β = .32, p < .001), which in turn, were associated significantly with defensive relational schemas (β = .15, p < .01; β = .19, p < .05, respectively). Defensive schema increased significantly the odds of being included in the Uncertain trajectory (OR = 1.74, p < .05) and Conflictual trajectory (OR = 3.13, p < .001). We also observed direct associations between socioeconomic instability and community disadvantage with relationship trajectory groups: socioeconomic instability significantly predicted membership in the Uncertain trajectory compared to the Normative trajectory (OR = 1.32, p < .05) while community disadvantage significantly predicted the membership in the Conflictual trajectory compared to the Normative trajectory (OR = 1.51, p < .05).

Figure 3.

Associations between early and contemporaneous contextual risk factors and romantic relationship trajectory classes in young adulthood

Note. N = 331. COD = Community disorder; CRS = Community resources (reverse coded); CVR = Cynical views of relationships; ECR = Experiences in close relationships; SCO = Street code. Reference group is Normative trajectory class. Nonsignificant pathways are not pictured. Pathways predicting defensive relational schemas are standardized betas; pathways to relationship trajectory classes (Uncertain and Conflictual) are standardized betas and odds ratios (in parentheses). Respondents’ age and having the same main partnership across waves 1 to 3 were controlled. *p < .05, **p < .01, ***p < .001

Significant indirect effects (see Supplemental Table 3) linked adverse childhood experiences to the Conflictual trajectory via socioeconomic instability and defensive relational schemas (b [95% CI] = .04 [.003, .076], p < .05) and via community disadvantage and defensive relational schemas (b [95% CI] = .06 [.007, .120], p < .05). Significant indirect effects also linked adverse childhood experiences to the Uncertain trajectory via socioeconomic instability and defensive relational schema (b [95% CI] = .03 [.002, .056], p < .05).

Discussion

Developing close, supportive relationships with a romantic partner during emerging adulthood is an important developmental task that may have long-term influences on psychological and physical health as well as engagement in marriage and marriage-like bonds in adulthood (Arnett, 2007; Fincham & Cui, 2010). Few studies have investigated romantic relationship development among African American men (Kelly & Floyd, 2006), and to our knowledge, this study provides among the first findings on development and change in rural African American men’s relationships during the emerging adult period. We also linked African American men’s exposure to adverse childhood experiences and contemporaneous contextual risk factors to how men’s relationship dynamics changed over time.

Using parallel process GMM analyses, we identified three trajectory-based profiles that represented changes in men’s relationship quality during the emerging adult years. Nearly half of the sample was classified as belonging to the Normative trajectory (48.9%), which started with low levels of conflict and high levels of support and trust at the initial assessment. Over time, their relationships exhibited decreases in conflict and increases in support and trust. This pattern is suggestive of normative or healthy relationship development and is consistent with developmental data which show that a majority of emerging adults feel positive about their relationships and report high levels of satisfaction (Wildsmith et al., 2013).

Although previous research has documented elevated conflict among African American couples (Kurdek, 2008), our findings suggest that nearly half of the African American young men in our sample view their relationships as increasingly supportive and well-functioning. A sizeable proportion of well-functioning relationships is particularly significant given the socioeconomic hardships many men experience and the elevated rates of adverse childhood experiences in the sample. It is also noteworthy that over one-third of these young men remained in romantic relationships with the same partners, and the average duration of their partnerships was two and a half years. Our findings show that among African American young men in our sample, positive relationship development is common, and suggest the importance of future research that investigates the resilience mechanisms that account for these positive outcomes.

The next largest group was the Uncertain trajectory (38.4%). They exhibited the lowest levels of support and trust at baseline; these qualities diminished over time. This group, compared to the others, evinced modest levels of relationship conflict over time. Even though men in this trajectory class did not perceive their relationships as supportive or particularly trust their partner, conflict remained modest. This pattern suggests either a relationship of convenience or one with low levels of commitment. As past research documented, emerging adults often become involved in romantic relationships with little expectation for long-term stability, suggesting that indifference or low levels of involvement might be common experiences (Halpern-Meekin, Manning, Giordano & Longmore, 2013).

A small portion of men were members of a Conflictual trajectory group (12.7%), characterized by high levels of conflict at baseline, which substantially increased over time. Young men in this group started with relatively low levels of dyadic trust, which further decreased over time. Although men in this trajectory group exhibited high levels of relationship support at the initial level, it also decreased substantially over time. The widening gap between relationship conflict and support/trust of this profile group indicates that young men in this group may experience repetitive, chronic relationship conflict. Prolonged conflict experiences are linked to the deterioration of psychological and physical well-being across the life course (Umberson et al., 2010). An escalating trajectory of conflict and decreasing trust also may forecast downstream problems with intimate partner violence (Schumacher & Leonard, 2005).

Study findings include developmental pathways linking childhood and contemporaneous risk factors to relationship trajectories. Consistent with our expectations, we found that adverse childhood experiences were linked to socioeconomic instability and community disadvantage, suggesting that rural African American young men who experienced childhood adversity were more likely to experience downstream exposure to stressors during emerging adulthood. This is consistent with life course perspectives that suggest negative life experiences, beginning in childhood, accumulate throughout the life course contributing to continued disadvantage during emerging adulthood (Umberson et al., 2014). Stability in stressful life experiences is thus initiated by early adversity which forecasts proximal stressors in adulthood (Pearlin et al., 2005). We also found that both childhood adversity and contemporaneous contextual risk factors were associated with men’s defensive relational schema. This is consistent with previous research demonstrating that exposure to harsh and stressful life contexts promotes mistrust of others, including romantic partners (Simons et al., 2012). The direct association between adverse childhood experiences and defensive relational schema indicates the enduring influence of childhood adversity on negative schemas. Defensive relational schema, although relatively enduring, may also be malleable, as suggested by associations with contemporaneous stressors in our study. This is consistent with past research indicating that contemporaneous stressors can modify schemas for better or worse (Kogan et al., 2013; Toth et al., 2002).

Defensive relational schemas were associated with membership in Conflictual and Uncertain trajectories. Young men with defensive relational schemas are likely to perceive romantic partners’ behaviors as untrustworthy, hostile, and problematic (Simons et al., 2014), which can contribute directly to the development of unstable and less satisfying intimate relationships. The longitudinal associations from childhood adversity to Conflictual and Uncertain trajectories via defensive relational schema show that the uncertainty stemming from difficult environments at earlier life stages spills over to romantic relationships (Burton & Tucker, 2009). Interestingly, socioeconomic instability and community disadvantage were associated with relationship trajectory class independent of relational schemas. As mentioned above, socioeconomic instability exhibited a direct effect to the Uncertain trajectory group. According to social bonding theory (Hirschi, 1969), exposure to disadvantaged environments erodes bonds to conventional institutions and participation in prosocial relationships. Indifferent and unsupportive involvement in romantic relationships, therefore, may reflect these young men’s difficulties or reticence to form strong bonds with their romantic partners. This might be, in part, due to their lack of personal and economic resources, which is linked to a willingness to invest in intimate relationships (Barr & Simons, 2012). Community disadvantage, on the other hand, directly predicted membership in the Conflictual trajectory group. General-strain theory (Agnew, 1992) argues that disadvantaged communities may promote norms and values that are conducive to deviant and hostile behaviors. Rural African American men’s relationship behavior may be influenced by the observation of troubled relationship behavior in their communities. At the same time, hostile and conflictual interactions between romantic partners may be considered acceptable in disadvantaged communities due to the erosion of community norms that deter aggressive interactions (Wilson, 2012).

Several limitations in the present study are noteworthy. The measure of adverse childhood experiences was a retrospective self-report and therefore could be subject to recall bias. In addition, our data assessed romantic relationships at the time point when the survey was conducted, and thus, we do not have information on all romantic relationships that respondents were involved in. Considering that cumulative relationship experiences (e.g., total number of romantic partners) may influence relationship quality, future research would benefit from including information on cumulative romantic experiences. Also, our study focused on changes in relationship quality over time. This focus may miss important information regarding the effects of transitioning in and out of relationships. Designs that incorporate both relationship quality and status in future research are needed. Given that our three trajectory classes are sample-dependent, our findings also should be interpreted with caution and need to be replicated with different samples. Our subsample GMM analysis, which included only respondents who had same partners across time, found almost identical trajectory classes to the main study. Although this supports the robustness of our findings in terms of changing relationships, we were unable to explore the antecedents of the classes in this auxiliary analysis due to statistical power limitations. Lastly, the current study is based on a rural sample, which may not be generalizable to urban couples. These limitations notwithstanding, the present study revealed heterogeneous romantic relationship trajectories that provide valuable information on rural African American young men’s relationship development. Our findings further support the contention that the conjoint influences of early adversity and contemporaneous stressors are robust antecedents of African American men’s romantic relationship behaviors during emerging adulthood.