Romantic Relationship Development: The Interplay Between Age and Relationship Length

Abstract

The present study explored how romantic relationship qualities develop with age and relationship length. Eight waves of data on romantic relationships were collected over 10.5 years during adolescence and early adulthood from a community-based sample in a Western U.S. city (100 males, 100 females; M age Wave 1 = 15.83). Measures of support, negative interactions, control, and jealousy were derived from interviews and questionnaire measures. Using multilevel modeling, main effects of age were found for jealousy, and main effects of relationship length were found for each quality. However, main effects were qualified by significant age by length interactions for each and every relationship quality. Short relationships increased in support with age. In comparison, long-term adolescent relationships were notable in that they were both supportive and turbulent, with elevated levels of support, negative interactions, control, and jealousy. With age, long-term relationships continued to have high levels of support, but decreased in negative interactions, control, and jealousy. Present findings highlight how the interplay between age and relationship length is key for understanding the development of romantic relationships.

Romantic relationships change significantly with age, from the relatively fleeting and casual experiences characteristic of adolescence to the more lasting and intimate bonds representative of adulthood. The typical transition between adolescent dating and enduring adult relationships may span over a decade of one’s life (Meier & Allen, 2009). Across this time, the qualities of romantic relationships shift, as individuals mature and gain experience across multiple relationships (Furman & Buhrmester, 1992; Giordano, Manning, Longmore, & Flanigan, 2009). Traditionally, the field has explored romantic relationship development by focusing on how relationships change with age. However, relationships also grow longer between adolescence and adulthood (Seiffge-Krenke, 2003), and longer relationships differ in their qualities from shorter ones (Connolly & Johnson, 1996). In effect, romantic relationship development occurs as a function of both age and relationship length, although very little is known about the interplay between age and relationship length. The purpose of the present study was to examine how qualities of romantic relationships change with age, relationship length, and the interaction between the two.

Changes in Romantic Relationships with Age

Multiple theories of relationship development posit that a sequence generally occurs in the nature of romantic relationships from adolescence to adulthood (e.g., Brown, 1999; Connolly & Goldberg, 1999; Furman & Wehner, 1997). In early adolescence, romantic relationships arise from mixed gender peer groups and tend to involve group dates, whereas in middle adolescence, dyads begin to be established, and by late adolescence, relationships start to resemble adult romantic relationships (Brown, 1999). One theory, the behavioral systems theory, proposes that as youth grow older, romantic partners increasingly become the central figure in the affiliative and sexual behavior systems, and ultimately, the attachment and caregiving systems (Furman & Wehner, 1994, 1997).

Moreover, Laursen and Jensen-Campbell’s (1999) developmental application of social exchange theory proposes that adolescents are focused on the self, and make relationship decisions predominately driven by personal gain. With age, individuals are thought to become less interested in maximizing personal rewards and grow focused on enhancing mutual gains by having both themselves and their romantic partner benefit in interactions.

These theories of relationship development suggest that as youth grow older, changes may occur in a number of relationship qualities. Four kinds of qualities have been conceptualized as key qualities in all close relationships: (a) positive qualities, as reflected by support, (b) negative qualities, as seen in negative interactions, (c) power/status arrangements, as manifested in controlling behaviors, and (d) comparisons with other relationships, as reflected by jealousy (Adler & Furman, 1988). Between adolescence and adulthood, romantic relationships increase in support and intimacy (Furman & Buhrmester, 1992; Seiffge-Krenke, 2003; Shulman & Kipnis, 2001), as well as other positive qualities such as satisfaction (Masked Citation 1)¹ and passionate love (Giordano et al., 2009). Young adults also describe their adolescent relationships as being more problematic than their current relationships (Shulman & Kipnis, 2001). However, existing empirical studies provide ambiguous findings for how negative qualities change with age, as conflict has been found to both decrease (Robins, Caspi, & Moffit, 2002) and increase with age (Masked Citation 2). With regards to changes in power, the amount of influence the romantic partner has on one’s behavior has been shown to increase with age: both the sheer number of attempts to influence one’s romantic partner and the actual perceived influence of the romantic partner increase from adolescence to adulthood (Giordano et al., 2009). Finally, jealousy in romantic relationships does not appear to change across the high school years (Seiffge-Krenke & Burk, 2013).

Aside from these studies, little is known about age changes in romantic relationships in adolescence and early adulthood. Furthermore, the literature on age changes is markedly divided. Existing studies either end at the cusp of adulthood (i.e., ages 19–22; Furman & Buhrmester, 1992; Giordano et al., 2009; Seiffge-Krenke, 2003), or studies start in early adulthood (i.e., age 18) and do not capture changes in adolescence (Robins, et al., 2002). Research is needed that spans both adolescence and adulthood to understand the full extent of changes that are occurring across this time.

Changes in Romantic Relationships with Length

Relationship length also changes between adolescence and adulthood. As noted previously, theories of relationship development delineate the typical sequence of relationship experiences as progressing from involvement in multiple, brief romantic relationships in adolescence to engagement in a single intimate relationship of longer length in early adulthood (Brown, 1999; Connolly & Goldberg, 1999). Consistent with this idea, older adolescents are more likely than younger adolescents to report relationships longer than eleven months (Connolly & Johnson, 1996). Relationships continue to grow longer into early adulthood, with romantic relationships at age 21 lasting almost four times longer than relationships at age 15 (Seiffge-Krenke, 2003).

According to social-exchange theory, the shift from egocentric motivations towards emphasizing dyadic gain is also expected to occur as relationships become longer (Laursen & Jensen-Campell, 1999). For example, when individuals are acquaintances, they tend to be motivated to maximize individual outcomes; when relationships become more interdependent, individuals shift toward optimizing mutual benefits. Thus, as length increases, a parallel shift towards mutual gains should occur.

Consistent with these theories, relationship length is linked to the quality of the relationship. Longer relationships have greater levels of support and with length, the romantic partner increases in salience as a support provider. Indeed, adolescents in longer relationships rated their romantic partner as more supportive than parents and friends, whereas individuals in shorter relationships rated their best friends and parents as more supportive than their romantic partners (Connolly & Johnson, 1996). Longer relationships also have more frequent daily interactions and greater interdependence (Adams, Laursen, & Wilder, 2001; Connolly & Johnson, 1996).

Although longer relationships have more positive interactions than shorter relationships (Connolly & Johnson, 1996; Rostosky, Galliher, Welsh, & Kawaguchi, 2000), studies also demonstrate that conflict increases with relationship length (Stafford, Kline, & Rankin, 2004). In comparison, findings regarding the link between length and power are unclear: longer relationships have been associated with more partner influence in one study (Giordano et al., 2009), but not with partner influence in another study (Adams, Laursen, & Wilder, 2001). Finally, longer relationships are associated with greater jealousy in romantic relationships during late adolescence (Seiffge-Krenke & Burk, 2013).

Age and Length and their Interaction

In sum, the qualities of romantic relationships are associated with both age and length. Age and relationship length may capture different facets of developmental time. Age reflects maturity and increasing experience with romantic relationships overall, whereas romantic relationship length captures the development of a relationship with a particular partner over time (Zimmer-Gembeck & Ducat, 2010). As such, it is important to understand the unique contributions each makes in romantic relationship development. Unfortunately, studies of age changes in romantic relationships have typically examined how relationship qualities are associated with either age or with length, but not both. One study of age-related changes controlled for length by including it as a covariate (Giordano et al., 2009), but to the best of our knowledge, no other studies have examined how both age and length are associated with qualities.

It is particularly important to examine both age and relationship length because they co-vary: as individuals grow older, their relationships tend to grow longer (Seiffge-Krenke, 2003). Consequently, a number of the age changes observed in romantic relationships may actually reflect the parallel increases in the length of relationships that happens with age. For example, the fact that young adults’ romantic relationships are more supportive than adolescents’ relationships may reflect the fact that young adults’ relationships are typically longer.

Not only is it important to simultaneously examine the unique roles of age and of length, but it is also important to examine their interplay—i.e., the interaction between age and length. Behavioral systems theory highlights that such a pivotal interaction emerges in late adolescence. In particular, the theory predicts that it is not until a longer relationship develops in late adolescence or early adulthood that the romantic partner assumes a central role in all four behavioral systems (Furman & Wehner, 1997). Once the partner has assumed this central role, such long-term relationships are expected to be particularly supportive—in fact, more supportive than would be expected than by examining the main effects of age and length. To the best of our knowledge, no studies have considered how age and relationship length might interact to shape relationship qualities.

Gender

Studies have also begun to demonstrate that males and females differ in their experiences of romantic relationships between adolescence and adulthood. Across ages, females report higher levels of support from their romantic partners than males do (Seiffge-Krenke, 2003). With regards to gender differences in power or control, findings are ambiguous: existing studies suggest that adolescent romantic relationships are egalitarian (Galliher, Rostosky, Welsh, & Kawaguchi, 1999), whereas other studies find that females are more likely to be perceived as being in power across all ages (Giordano et al., 2009), and yet other studies suggest that in late adolescence and early adulthood, males have more power (Furman & Buhrmester, 1992). Finally, with regard to gender differences in jealousy, females report higher levels of jealousy in romantic relationships during high school (Seiffge-Krenke & Burk, 2013).

The Present Study

The present study used a longitudinal design to examine how the qualities of romantic relationships vary as a function of age, length, the interaction of age and length, and gender. Previous longitudinal studies have either ended at the beginning of adulthood (Furman & Buhrmester, 1992; Giordano et al., 2009; Seiffge-Krenke, 2003), or started in early adulthood (Robins, et al., 2002). In order to capture the changes that occur in relationships across adolescence and adulthood, eight waves of data spanning ages 15–25 were included. We explored four relationship qualities--support, negative interactions, control, and jealousy. Whereas existing literature on romantic relationship development has relied on self-report questionnaires, the present study included both self-report and interview data on the qualities of relationships.

One of the primary purposes of our study was to examine both age and length simultaneously to disentangle each variable’s unique contribution to romantic relationships. We examined the association between romantic relationship qualities and age while controlling for the length of the relationship, and examined the association between qualities and length while controlling for the participants’ age. Finally, the present study examined how the interaction between age and relationship length shapes romantic qualities, a question that has not been previously explored.

Hypotheses

Consistent with development theory and existing literature, we hypothesized that relationships would increase in support with age (H1). We also anticipated that longer relationships would be more supportive than shorter ones (H2). In comparison, predictions regarding negative interactions are less apparent; on the one hand, one might expect increases in support in romantic relationships to be complemented by a decrease in negative interactions. On the other hand, as relationships become more interdependent, opportunities for conflict may increase. As such, no hypotheses were garnered for how negative interactions might change with age and relationship length.

Furthermore, with age and relationship length, individuals are expected to shift from focusing primarily on the self to increasingly focus on the relationship and on mutual gain. Consequently, controlling behaviors, which are driven by maximizing personal rewards, were hypothesized to decrease with age and length (H3 & H4). Increases in mutuality with age and with length were also expected to lead to decreases in feelings of jealousy, as these increases in mutuality would promote security within the relationship (H5).

Regarding the interplay between age and relationship length, it was thought that the role of relationship length would vary with age. Specifically, it was anticipated that the hypothesized length effects would be greatest in adult relationships (H7). For example, we hypothesized that increases in support with length may be most striking in adulthood, when there has been more time for the romantic partner to become the primary person to turn to. Similar considerations would also lead one to expect multiplicative effects of age and length for decreases in controlling behaviors and jealousy.

Finally, we expected that females would have higher levels of support than males, as females report more emotional intimacy in their relationships (Shulman & Scharf, 2000). We did not garner any hypotheses about gender and conflict, as research has not found any gender differences (Robins, et al., 2002). We anticipated that females would have higher levels of jealousy overall (Seiffge-Krenke & Burke, 2013). Finally, previous studies have mixed findings regarding gender and control or power (Galliher, Kawaguchi, Rostosky, & Welsh, 1999; Giordano, et al., 2009). As such, we did not make any specific hypotheses about gender differences in control.

Method

Participants

Data were drawn from a large longitudinal study of close relationships and psychosocial adjustment across adolescence and young adulthood. Two hundred adolescents in the 10^th grade (100 males; 100 females) were initially recruited from a Western metropolitan area. In order to obtain a diverse sample, brochures were distributed to students enrolled in a number of schools in ethnically diverse neighborhoods, and letters were mailed to families across a variety of zip codes. Because we used brochures and sent letters to many families who did not have a 10th grader, we were unable to determine the ascertainment rate. Interested families were contacted with the objective of selecting a quota sample with a distribution of racial/ethnic groups that approximated that of the United States and had an equal number of males and females. Families were compensated $25 to hear a description of the project in their home to insure maximal response; 85.5% of families expressed interest and carried through with the first assessment.

Of the sample, 69.5% identified as White, non-Hispanics, 11.5% identified as African Americans, 12.5% identified as Hispanics, 1.5% identified as Native Americans, 1% identified as Asian American, and 4% identified as biracial. 57.5% of the participants resided with two biological or adoptive parents, 11.5% resided with a biological or adoptive parent and a stepparent or partner, and the remaining 31% resided with a single parent or relative. The sample was of average intelligence (WISC-III vocabulary score M = 9.8, SD = 2.44); 55.4% of their mothers had a college degree, indicating that the sample was predominately middle or upper middle class.

In terms of sexual orientation, 90.5% reported they were heterosexual/straight at Wave 8, whereas the other participants reported they were bisexual (3%), gay (3%), lesbian (1.2%), or questioning (2.4%). We retained the sexual minorities in our sample to be inclusive and because we had no theoretical or empirical rationale for believing that the patterns of associations would differ by sexual orientation.

As part of the larger project, participants and their mothers completed several questionnaires in Wave 1. Although most of these measures are not directly relevant to the present study, we compared our sample’s scores to comparable national norms of representative samples for trait anxiety scores on the State Trait Anxiety Inventory (Spielberger, 1983), maternal report of externalizing symptoms on the Child Behavior Child Checklist (Achenbach, 1991), participants’ reports of internalizing and externalizing symptoms on the Youth Self Report, and a number of indices of substance use from the Monitoring the Future survey (Johnston, O’Malley, & Bachman, 2002). The present sample was more likely to have tried marijuana (54% vs. 40%, z = 2.23, p < .05); however, the sample did not differ significantly from national scores on the other 11 measures, including frequency of marijuana usage.

Procedure

For the present study, data were drawn from Waves 1 through 8 of the study (Wave 1 mean age = 15 years, 10.44 months, SD = 0.49, range 14–16 years old; Wave 8 mean age = 25 years, 7.99 months, range = 24–27 years). The eight waves of data were collected between 2000 and 2012. In each wave, participants completed questionnaires and were individually interviewed about their romantic relationships. During Waves 1–4, data were collected on a yearly basis. During Waves 5–8, data were collected every eighteen months. Participant retention was excellent (Wave 1 & 2: N = 200; Wave 3: N= 199; Wave 4: N= 195; Wave 5: N= 186; Wave 6: N= 185; Wave 7: N= 179; Wave 8: N = 174).

Those who participated in the study in Wave 8 did not differ from those who did not in terms of age, ethnicity, and maternal education. Furthermore, we compared those who did not participate in Wave 8 to those who did participate on the length and the qualities of their relationships in Waves 1, 2, 3, & 4 when almost all were in the study. Out of the twenty comparisons, there was only one significant difference, which is what would be expected by chance. In Wave 4, those who did not participate in Wave 8 had lower support in their romantic relationships than those who did participate. We also compared the number of romantic relationships across all previous waves for the individuals who did not participate in Wave 8 to those who remained in the study. Again, there were no differences: those who did not participate in Wave 8 had as many relationships as those who had remained in the study.

The study was approved by the local Institutional Review Board. The confidentiality of participants’ data was protected by a Certificate of Confidentiality issued by the U.S. Department of Health and Human Services.

Measures

Demographic Information

In each wave, participants reported the length of their romantic relationship in months for the most important romantic relationship in the past year that had lasted one month or longer duration. This is the partner who they completed self-report measures about and were interviewed about. Participants also reported whether they were cohabiting with their partner. Finally, age was calculated by subtracting the participants’ birthdate from the date when they were interviewed.

Network of Relationships Inventory

Participants completed the Network of Relationships Inventory: Behavioral Systems Version (NRI; Furman & Buhrmester, 2009) about their most important romantic relationship in the last year. Participants completed five items regarding social support (e.g., “How much do you turn to this person for comfort and support when you are troubled about something?”) and six items regarding negative interactions, conflict, and antagonism (e.g., “How much do you and this person get on each other’s nerves?”). Participants used a five-point scale to rate how characteristic each description was of their relationship. Support and negative interaction scores were derived by averaging the relevant items (M α = .89 & M α = .92, respectively).

Multidimensional Jealousy Scale

Jealousy was measured using Pfeiffer and Wong’s (1989) Multidimensional Jealousy Scale. The measure included questions about emotional jealousy, cognitive jealousy (e.g. how often one is suspicious about their partner becoming interested in someone else), and behavioral jealousy (e.g., asking about the partner’s whereabouts). Participants completed 24 questions using a five-point Likert scale (M alpha = .91). The 24 items were averaged to derive a total score.

Romantic Interview

The Romantic Interview (RI; Furman, 2001) was used to assess participants’ interactions within romantic relationships. The RI was based on the Adult Attachment Interview (George, Kaplan, & Main, 1985/1996). Many questions were the same or similar to those of the AAI. For example, participants were asked to describe their romantic relationships using specific memories to support descriptions. They were asked about separation, rejection, threatening behaviors, and being upset within their romantic relationship.

For the present study, descriptions from the most important romantic relationship in the past year were used. The RIs were audiotaped and transcribed verbatim. Crowell and Owen’s (1996) Current Relationship Inventory (CRI) coding system was used to rate relationship qualities reflecting the participants’ support seeking and providing, conflict, participants’ controlling behaviors, and participants’ involving behaviors. All coders attended Main and Hesse’s AAI Workshop and received additional training in coding the Romantic Interviews. Reliabilities of the different relationship qualities were satisfactory (M intraclass correlation coefficient (ICC) = .69, Range = .57 to .78).

Interview rating of support

Coders separately rated support seeking and support providing by the participant. Support seeking refers to expressing distress, accepting comfort, and using the other as a secure base. Support providing refers to providing support at times of distress and serving as a secure base for one’s romantic partner. The scores of the two scales were averaged to derive a support composite.

Interview rating of negative interactions

Coders rated the amount of conflict in the romantic relationship, taking into account its intensity and frequency. Interview ratings of conflict were conceptualized as analogous to the self-report of negative interactions, but from interview ratings.

Interview rating of control

Coders also rated the participant’s level of control in the romantic relationship. Controlling behavior was scored when one person exercised power over the other and imposed his or her ideas on the other partner in order to get his or her way.

Interview rating of involving behaviors

Coders rated the participant’s involving behaviors, or behaviors designed to keep the other focused on them and the romantic relationship. Involving behaviors are those which heighten attention, affection, and closeness within the romantic relationship by limiting the other person’s exploration and autonomy, limiting the other person’s feelings of confidence, and keeping the romantic partner focused on oneself. Involving behaviors also include expressions of sexual jealousy and attempts to make the partner jealous and were conceptualized as analogous to the self-report of jealousy, but based on interview ratings.

Composites

As the interview and self-report scales were substantially correlated with one another (M r for support = .41, M r for negative interactions = .50, M r for jealousy = .41), they were combined into composites. The various measures used to create the composites had different numbers of points on their scales, which presents problems in deriving a composite as the scores are not comparable; consequently scale scores were standardized across all waves to render the scales comparable with one another, a recommended procedure that retains differences in means and variance across age, and does not change the shape of the distribution or the associations among the variables (Little, 2013). Standardized scores on the self-report and interview measures were then averaged to form the composite.

Results

Preliminary and Descriptive Analyses

All variables were examined to insure that they had acceptable levels of skew and kurtosis (Behrens, 1997). Outliers were Winsorized to fall 1.5 times the interquartile range below the 25^th percentile or above the 75^th percentile. Additional descriptive statistics can be found in Table 1. In Wave 1, 59.8% of participants reported having had a romantic partner in the past year, whereas in Wave 8, 78.2% reported having had a romantic partner (see Table 1 for N’s in each wave). When participants did not have a romantic relationship in a particular wave, relationship qualities were missing. Only participants who reported having a romantic partner in at least one of the waves were included in analyses. Accordingly, 2.0% of participants were excluded.

Table 1.

Mean (standard deviation) relationship length and romantic qualities

	Wave 1	Wave 2	Wave 3	Wave 4	Wave 5	Wave 6	Wave 7	Wave 8
N participants in a relationship (N = 1055)	108	124	142	141	134	141	136	129
Age (N = 1040)	15.88 (0.47)	16.89 (0.47)	17.95 (0.50)	19.04 (0.56)	20.52 (0.56)	22.12 (0.51)	23.71 (0.60)	25.67 (0.62)
Relationship Length (N = 1055)	4.54 (3.60)	6.99 (5.42)	9.73 (8.11)	12.39 (11.45)	16.00 (14.03)	20.58 (20.01)	27.18 (24.52)	37.54 (32.87)
Support (N = 1050)	−0.45 (0.86)	−0.29 (0.89)	−0.25 (0.85)	−0.20 (0.86)	−0.15 (0.91)	−0.07 (0.92)	−0.06 (0.92)	−0.01 (0.92)
Negative Interactions (N = 1046)	−0.12 (0.80)	−0.13 (0.80)	0.10 (1.11)	−0.08 (0.90)	0.10 (0.94)	0.05 (0.90)	0.04 (0.78)	0.01 (0.87)
Jealousy (N = 1044)	0.06 (0.85)	−0.04 (0.81)	0.15 (0.87)	0.03(0.87)	0.11 (0.92)	0.03 (0.89)	−0.08 (0.76)	−0.26 (0.92)
Control (N = 930)	−0.08 (0.89)	−0.00 (1.05)	−0.00 (1.03)	−0.01 (1.03)	0.10 (1.13)	0.12 (1.15)	0.02 (0.90)	−0.14 (0.75)

Note. N’s in the far left column indicate the cumulative numbers across waves for that particular variable.

Age and length of the relationship were correlated across the eight waves (r= .49, p < .001). The mean relationship length increased with age (see Table 1). To ascertain whether the correlation between age and length was the same at younger and older ages, we divided our dataset into two groups based on the age of the participants. The correlation between age and length in participants younger than the median age of the sample (19.35 years old) was almost identical to the correlation between age and length for participants older than the median age of the sample (r= .35, p < .001 & r= .32, p < .001, respectively). These correlations suggest that there is substantial variability in relationship length throughout this age range.

Developmental Changes in the Qualities of Romantic Relationships

To test hypotheses, a series of multilevel models were conducted using the statistical program Hierarchical Linear Modeling (HLM Version 6.0; Raudenbush, Bryk, & Congdon, 2004). HLM takes into account the nested nature of the data in a longitudinal study. The models had the following form:

Level 1:	Yti = β0ti + β1ti (Cohabiting) + β2ti (Past/Present Relationship)+ β3ti (Age) + β4ti (Length) + β5ti (Age x Length) + Rti
Level 2:	β0i = γ00 + γ01(gender) + U0i β1i = γ10 β2i = γ20 β3i = γ30 β4i = γ40 β5i = γ50

In these models, Y_ti represented the relationship quality at time t for individual i. The participant’s relationship status (not cohabiting versus cohabiting; higher scores indicate cohabitation) was included as a control variable to ensure that the changes in qualities that happen with age and relationship length were happening beyond changes in relationship status. Additionally, the participant’s report on either a present or past relationship was included as a control variable (β₂ past/present relationship; higher scores indicate present relationships).

We used a hierarchical model to examine associations, with both age and relationship length grand mean centered. The significance level was adjusted for false discovery rates (Benjamini & Hochberg, 1995). First, we conducted a model with age in years (β₃), relationship length in months (β₄), and gender (γ₀₁). We entered the interaction effects after the main effects to avoid the limitations of interpreting conditional main effects (Cohen, Cohen, West, & Aiken, 2003; Little, 2013). The main effects and interactions are presented together in Table 2; however, the unstandardized regression coefficients and standard errors for the main effects and interactions are the values from the respective step at which they were entered in the analyses. In preliminary analyses, interactions between gender and length or age were included; only 1 of 12 effects was significant, and thus, these interactions were not included in the primary analyses.

Table 2.

Multilevel Models Testing the Associations between Romantic Relationship Qualities and Age, Length, and the Interaction Between Age and Length

	Support	Negative Interactions	Control	Jealousy
Intercept (β0)	−0.15(0.16)	−0.02(0.05)	0.11(0.06)	0.01(0.05)
Cohabitation (β1)	0.16†(0.09)	−0.18†(0.11)	0.07(0.14)	−0.48***(0.11)
Past/Present Relationship (β2)	0.61***(0.05)	−0.07(0.06)	0.06(0.07)	−0.04(0.05)
Age (β3)	0.01(0.01)	−0.01(0.01)	−0.01(0.01)	−0.04***(0.01)
Relationship Length (β4)	0.01***(0.00)	0.01***(0.00)	0.00*(0.00)	0.01***(0.00)
Age X Length (β4)	−0.00***(0.00)	−0.00***(0.00)	−0.00***(0.00)	−0.00***(0.00)
Gender Main Effect (γ01)	0.24***(0.05)	−0.08(0.08)	0.08(0.09)	0.22**(0.08)

Notes.

^†p < .10;

^*p < .05;

^**p < .01;

^***p < .001.

The primary numbers in the table are the unstandardized coefficients.

Main Effects of Gender, Age, and Length

Significant effects of gender were found for two relationship qualities. Females were higher in levels of support and jealousy. Significant main effects of age were found for jealousy, which decreased with age. Support, negative interactions, and control did not change with age. Significant main effects of length were found for all variables. Consistent with hypotheses, ratings of support increased with relationship length. Additionally, negative interactions, control, and jealousy increased with relationship length.

Interaction between Age and Relationship Length

All main effects of age and of relationship length were qualified by interactions between age and length. To further interpret our significant interactions, we used Preacher, Curran, and Bauer’s (2006) computational tools to probe interactions. The values of the main effects used in calculating these graphs were based on the full interaction model. A table including these conditional values of the main effects is available online. For the age effects, we plotted the estimated effects of age on romantic qualities for three lengths: short (4 months), medium (9 months), and long-term (22 months). At all ages, participants had short, medium, and long relationships. As such, we chose these lengths as meaningful indices of the amount of time spent in a romantic relationship (4 months – 25^th percentile; 9 months – 50^th percentile; 22 months – 75^th percentile).

For support, short relationships increased in levels of support with age (see Figure 1), B = 0.02, t(1015) = 2.41, p <.05. Medium length and long-term relationships did not change in levels of support with age.

Figure 1.

Interaction between age and relationship length on ratings of support.

Short and medium length relationships did not change in levels of negative interactions with age. However, negative interactions decreased in long-term relationships with age (see Figure 2), B = −0.05, t(1015) = −4.65, p <.01.

Figure 2.

Interaction between age and relationship length on ratings of negative interactions.

Jealousy tended to decrease in short relationships with age and decreased significantly in medium and long-term relationships with age; the age effects were strongest in the long-term relationships (see Figure 3), B = −0.02, t(1012) = −1.87, p = .06; B = −0.03, t(1012) = −3.11, p < .01; and B = −0.06, t(1012) = −6.12, p < .001, respectively.

Figure 3.

Interaction between age and relationship length on ratings of jealousy.

Control did not change with age in short or medium length relationships (see Figure 4). Only long-term relationships decreased in control with age, B = −0.05, t(907) = −3.54, p < .001.

Figure 4.

Interaction between age and relationship length on ratings of control.

As a sensitivity test, we also examined the effects of age on romantic qualities for two additional lengths: 5 months (33^rd percentile) and 17 months (66^th percentile). These results match the results presented above in that 5-month relationships had the same effects as short relationships and 17-month long relationships had the same effects as long relationships (see supplementary analyses).

Relationship Length Effects by Age

Up to this point, we have described how age effects depended upon the length of the relationship. We also examined how length effects depended upon age. Once again, we used Preacher et al.’s (2006) computational tools to probe the interactions. For the length effects, we plotted the estimate effects of age on romantic qualities at three ages: age 17 (closest whole value to the 25^th percentile), age 20 (closest whole value to 50^th percentile), and age 23 (closest whole value to 75^th percentile). Notably, there was an effect of relationship length on all relationship qualities. Longer relationships had more support, but also had higher levels of negative interactions, control, and jealousy. These effects of length were stronger at younger ages, but significant at all three ages. Graphs of these interactions are presented as supplemental material.

Discussion

In adolescence and early adulthood, the nature and qualities of romantic relationships change significantly. The current study explored the relatively uncharted territory of how qualities unfold across this time, examining development both in terms of the youth’s age and the length of the relationship. Moreover, the study also contributed to our understanding of relationship development by spanning a wide range of ages across both adolescence and early adulthood, assessing an array of relationship qualities, and including interview ratings as well as self-reports.

Consistent with prior research (Seiffge-Krenke, 2003), age and length were found to be moderately associated with each other. Such covariation has made it difficult to disentangle the effects of age and length in past work, which has usually only examined one of these variables. By examining both of these effects simultaneously, we were able to determine that both age and length each uniquely contributed to romantic qualities. Jealousy decreased with age, but increased with length, further underscoring the distinct contribution of the two variables. For the other qualities, we only found a main effect of length and not a main effect of age. Such findings suggest that some of the seeming age effects found in prior research may actually reflect the effects of relationship length. These findings highlight the importance of examining both age and length in relationships during adolescence and young adulthood.

At the same time, all main effects were qualified by interactions between age and relationship length for each and every relationship quality. In effect, adult relationships did not consistently differ from adolescent ones; it depended on the length of the relationship. Similarly, longer relationships did not consistently differ from shorter relationships; it depended on the youth’s age. Thus, prior research has not captured the complexity of developmental changes that occur from adolescence to adulthood.

Our multifaceted findings are best understood by contrasting the developmental changes in short relationships to the developmental changes in long relationships; not surprisingly, changes in medium length relationships fell in between these two.

Developmental Changes in Short Relationships

In line with our predictions, short relationships increased in support from adolescence to early adulthood. This finding is consistent with behavioral systems theory, which posits that as individuals grow older, they increasingly turn to their romantic partner as a companion, an attachment figure, and a person to take care of (Furman & Wehner, 1994, 1997). With age, individuals may become adept at quickly establishing supportive exchanges due to the skills and experience accrued across previous relationships. Even in short relationships, young adults are able to promote supportive interchanges, perhaps in order to more quickly determine whether a partner could potentially become a long-term partner.

Developmental Changes in Long Term Relationships

The most striking change regarding long-term relationships is that they become far more common in early adulthood. Such long-term relationships may be less common among younger individuals because they may be less interested in having such a relationship or they may lack the skills to sustain a long-term relationship.

Although the long-term romantic relationships that do occur in adolescence are perceived to be just as supportive as those in adulthood, the levels of negative interaction, jealousy, and control in these relationships are also substantially higher than in early adulthood. In effect, adolescents’ long-term relationships appear to be rather turbulent.

Adolescents are likely to be in their first long-term relationship, and few of their peers have such relationships. Thus, they may perceive their long-term relationships to be very supportive because they are special and novel at that age. The romantic nature of these novel relationships may also make them feel closer, more intimate, and more supportive than other relationships.

At the same time, adolescents are motivated to maximize their own personal gain and tend to be more selfish, which would likely impact the quality of their relationships (Laursen & Jensen-Campbell, 1999). Adolescents’ egocentrism and focus on personal gain could lead to heightened levels of negative interactions as individuals try to get their own way and are less willing to compromise. Likewise, jealousy may emerge as a result of wanting their partner’s constant attention and being jealous of their partner’s attention to anyone else. Finally, control is another manifestation of maximizing personal gain, as an individual tries to influence their romantic partner’s behavior. It is also likely that adolescent romantic partners engage in behaviors that elicit jealousy or provoke conflict.

Thus, adolescent long-term relationships are intense in nature, and characterized by a mixture of both positive and problematic interchanges. Adolescents’ limited relationship skills, and the novelty and significance of these relationships may make them difficult to manage. With age, romantic relationships appear to become easier to handle. As youth become adults, they may feel more comfortable and secure in their long-term relationships and feel less jealous and controlling. In young adulthood, individuals also begin to emphasize maximizing mutual gains, such that both members of the dyad benefit (Laursen & Jensen-Campell, 1999). This shift toward mutual gains may unfold only at the intersection of older age and longer relationship length, when one is both cognitively mature and invested in the relationship. As such, the shift towards mutual gains likely parallels the decreasing levels of conflict, jealousy, and control with age in longer relationships.

Effects of Relationship Length at Different Ages

Up to this point, we have discussed how the qualities of short, medium, and long relationships change with age. We also examined how the effects of relationship length vary at age 17, 20, and 23. Longer relationships had more support, but also had higher levels of negative interactions, control, and jealousy. These effects were significant at all three ages, but notably, the effects at younger ages were stronger.

Taken together, conflict, control, and feelings of jealousy appear to be inherent as individuals become increasingly invested in the relationship, and as the relationship increases in interdependence. Although these dynamics do increase with length during young adulthood, it is not to the same degree that they do at earlier ages. Adolescents may find it more challenging to manage these specific romantic relationship dynamics. In comparison, young adults appear to be more skilled at handling these dynamics and perhaps have learned from previous experience. Young adults are also beginning to consider which qualities are important for selecting a life partner, and relationships that have elevated levels of these dynamics in earlier stages during young adulthood may end.

Interestingly, support did not increase as much with relationship length in young adulthood when compared to earlier ages. Overall, young adults may be able to elicit support in their romantic relationships regardless of the length of the relationship. As noted earlier, support increases even in short relationships from adolescence to young adulthood.

Implications for Clinical Work

Results from the present study have important applications for clinical work. Our findings provide a general guide for what qualities can be expected from short, medium, and long-term relationships in adolescence and adulthood. Those who work with adolescents and young adults should attend to both the age of the individual as well as the length of their relationship to understand what the relationship is likely to be like. For example, they should anticipate that long-term adolescent romantic relationships are tumultuous, which may be due to an egocentric tendency to focus on the self, a lack of competence and experience in navigating novel interpersonal dynamics, or low levels of confidence in one’s ability to interact with a romantic partner (Giordano et al., 2009). Furthermore, involvement in long-term romantic relationships may be particularly difficult for youth as they also face other developmental tasks such as establishing their identity (Connolly et al., 2014). Thus, these relationships will understandably be challenging for adolescents to manage, which is important for parents, educators, and clinicians to understand. Intervention efforts should focus on helping teenagers acquire the tools they need to manage these challenging dynamics and build healthy relationships.

Furthermore, parents, educators, clinicians, and even youth themselves often do not have a clear sense of what to expect as a relationship develops over time. It is important for adolescents and young adults to be told that feelings of jealousy, conflict, and control commonly increase as relationships grow longer. Accordingly, one of the challenges of maintaining a relationship is to cope with these changes that will likely occur as the relationship unfolds.

Processes Underlying Changes

In discussing our findings, we proposed a number of different processes that might underlie changes in relationship qualities. Drawing from behavioral systems theory (Furman & Wehner, 1994, 1997), we suggested that as individuals grow older and relationships grow longer, individuals may increasingly turn to their romantic partner as a person to affiliate with, as an attachment figure and as person to take care of. Second, relationship skills may develop with age and as a relationship grows longer. Finally, individuals may shift from egocentric motivations toward placing more emphasis on mutual relationship gains, which may lead to decreases in negative interactions, jealousy, and control in longer relationships.

If these conceptualizations prove to have merit, it would suggest that multiple processes may underlie changes in relationship qualities. Moreover, a single underlying process can influence multiple qualities of a relationship. For example, a shift toward mutual relationship gains could explain decreases in negative interactions as well as jealousy. Thus, it is important to examine the pattern of changes in relationships, and not just a single characteristic. At the same time, none of the hypothesized processes were directly measured in the present study. A valuable step for future research would be to measure such processes to determine if changes in these processes are indeed associated with changes in the qualities of relationships. It is hoped that our discussion of what might underlie our observed effects will stimulate such work and further theoretical development.

Gender

We found that males and females differ in their experiences of romantic relationship qualities. Consistent with existing literature (Seiffge-Krenke & Burk, 2013), females reported higher levels of jealousy. Research demonstrates that females tend to be more worried overall about losing the relationship and that they are more likely to interpret ambiguous scenarios in a jealous manner (Lans, Mosek, & Yagil, 2014). Future studies should explore more nuanced facets of jealousy, such as gender differences in the expression of jealousy.

In our study, females also reported higher support. Females tend towards a relational orientation (Maccoby, 1990) and experience more supportiveness in their friendships, which they may apply to their romantic relationships (Rose & Asher, 2004). Accordingly, females may be more comfortable seeking and providing support in a romantic relationship.

Heterogeneity of Relationships

Findings from the present study repeatedly underscore the complexity of development. A single normative pathway does not exist from adolescence to early adulthood, and changes do not uniformly unfold with age. Instead, our findings emphasize that relationships of varying lengths at different ages are remarkably heterogeneous. Short relationships in adolescence differ from short relationships in adulthood, but not in the same way that long relationships in adolescence differ from long relationships in adulthood. Our findings illustrate that a variety of relationships emerge in adolescence and adulthood, and help us begin to understand the quality and nature of these varied relationships. Research, theory, and clinical work will need to continue to take into account such heterogeneity to understand psychosocial development.

Limitations and Future Directions

The present study is one of the first to examine relationship development simultaneously in terms of age and relationship length. However, several limitations within the present study should be recognized. First, due to the size of our sample, power may have been limited in the present study, preventing us from detecting certain associations such as other potential main effects of age. Second, even though the study is longitudinal in nature, causal inferences cannot be made. In particular, we discussed how qualities may change as the relationship lengthens, but it is equally plausible that the quality of the relationship affected how long the relationship lasted. It is also possible that a third variable such as relationship-maintaining behaviors or beliefs could also explain the associations between length and the quality of the relationship. For example, implicit beliefs that a successful romantic relationship is characterized by growth and development are associated with both the longevity and the quality of the relationship (Knee, 1998). Likewise, rejection sensitivity could be associated with both the length of the relationship and the ensuing quality of the relationship: highly rejection sensitive individuals are more likely to have relationships that end, and have different rates of conflict in their romantic relationships (Downey, Freitas, Michaelis, & Khouri, 1998).

Although we have a rich, multiple method, longitudinal dataset with excellent retention (87% in Wave 8), we recognize that missing data is a limitation of the present study, as complete datasets are stronger than datasets with missing data in either predictor or outcome variables. Moreover, it is not possible to determine if data are Missing at Random (MAR). Indeed, other variables might be able to explain why data are missing. For example, although most participants had experienced a number of breakups (M = 11.89 by Wave 8), it is possible that some participants dropped out of the study after their relationship broke up.

In most instances, the youth were in different relationships at the different waves of data collection. It would be intriguing to focus on changes within the same relationship over time. By doing so, we could determine if the qualities of a relationship early on predict how long it lasts. Such a study could also rule out the possibility that the associations with relationship length are a reflection of the kind of person who has a long-term relationship versus those who have short-term relationships.

One of the more interesting findings in the study was that long-term adolescent romantic relationships seemed remarkably different from long-term adult relationships. Such relationships in the teenage years appear turbulent; although perceived as very supportive, these relationships were also characterized by high levels of negative interactions, jealousy, and control. More work is needed to understand the interpersonal dynamics in such relationships. In particular, the links between long-term adolescent romantic relationships and psychosocial development and adjustment merit further attention. In fact, a critical issue more broadly is the long-term mental health outcomes associated with relationships of different qualities and lengths at different ages.

Lastly, our study followed participants from late adolescence to early adulthood. It would also be informative to extend the study further into adulthood to understand how the qualities of the romantic relationship change as relationships grow even longer, and as greater numbers of individuals begin to form life partnerships, cohabit, and get married.

Although a number of key questions remain to be answered, the present study contributes to the literature on a phase of romantic relationship development that can span over a decade of one’s life. Previous studies have primarily explored relationship development by examining how relationships change with age. However, the results of our study show that age only captures a single facet of development and as such, only a part of the story of development has been understood. Notably, relationships are not static: they are growing and changing with time. The present study shows that we not only need to examine development across both age and relationship length, but that we also need to consider their interplay. Our findings consistently showed that the age changes differed as a function of the length of the relationship. Conversely, the qualities of a relationship of a particular length depended on an individual’s age. Thus, findings repeatedly underscored that the interplay between age and length are critical for understanding relationship development.