February 14, 2008
In Times of Uncertainty: Predicting the Survival of Long-Distance Relationships
By Cameron, Jessica J Ross, Michael
ABSTRACT. The authors examined the degree to which ratings of negative affectivity (NA) and relational security predicted the breakup of long-distance and same-city dating relationships. Couples completed initial surveys and were contacted 1 year later about the status of their relationship. In the initial surveys, both partners completed NA and relational security assessments. Overall, both the NA and relational security of men and women predicted stability. However, as predicted, structural equation modeling revealed a gender difference in the interaction between NA and long-distance status. The presence of high NA in men was associated with breakup for long-distance but not same-city couples. High NA in women was not differentially associated with relational stability on the basis of long-distance status. The authors discuss the psychological basis of this gender difference. Keywords: gender differences, negative affectivity, personality, relationships
LONG-DISTANCE RELATIONSHIPS are becoming increasingly common (Guldner, 2003). About one third of all dating relationships among university students are long distance (Stafford & Reske, 1990), and the proportion is even greater among 1st-year university students (Aylor, 2003). Physical separation of partners is associated with increased distress and depression (Guldner, 1996) and reductions in relationship satisfaction (Van Horn et al., 1997) and stability (Helgeson, 1994a; Lydon, Pierce, & O'Regan, 1997). Nonetheless, many long-distance dating relationships do survive. What are the conditions that lead some long-distance relationships to survive and others to fail?
Although few researchers have investigated the persistence of long-distance relationships, many have studied dating and marital stability. Most researchers interested in relational stability in proximate couples have examined characteristics of the relationship and of the individuals within those relationships as predictor variables (see Cate, Levin, & Richmond, 2002, and Karney & Bradbury, 1995, for reviews). Most relational stability research on distant daters has focused on the characteristics of physical separation, such as the number of miles separating residences (e.g., Carpenter & Knox, 1986) or the method of communication between partners (e.g., Dainton & Aylor, 2002), whereas few researchers have investigated characteristics of the relationships themselves (e.g., Schwebel, Dunn, Moss, & Renner, 1992). Researchers have also neglected to investigate the influence of individual differences on the stability of long-distance relationships. We argue that one personality characteristic-negative affectivity (NA)-is especially important for understanding stability in long-distance relationships. NA encompasses (a) dispositional pessimism about the future, (b) low self-esteem, and (c) the tendency to experience negative emotions such as anxiety and depression (Watson & Clark, 1984). High-NA individuals are less able to cope with general stress (Bolger & Zuckerman, 1995; Marco & Suls, 1993) and may therefore be particularly ill-suited to cope with the physical separation in long- distance relationships. Accordingly, the purpose of the present study was to examine whether NA differentially predicted the stability of long-distance and same-city relationships. To accomplish this goal, we followed dating couples involved in long- distance and same-city relationships over the course of 1 year to assess stability.
Predicting Relational Stability
Research has indicated that being involved in an enduring romantic relationship is beneficial for both physiological health and psychological well-being (Pierce, Sarason, & Sarason, 1996). Many psychologists have pursued ways to understand how people can establish stable relationships and consequently reap these benefits. Researchers investigating relational stability have focused on understanding the characteristics of stable relationships, the characteristics of individuals within those relationships, and the environments that facilitate or hinder these relationships (see Cate et al., 2002, for a review). Perhaps the most intuitive finding in this area is that people who are unhappy with their relationships are more inclined to break up (e.g., Karney & Bradbury, 1995; MacDonald & Ross, 1999). However, underlying this sense of relationship satisfaction rests a more important variable: relational security, or the sense of trust and faith one has in one's partner and relationship (Reis, Clark, & Holmes, 2004). Relational security appears to be a necessary requirement for progress in any interaction. Relationships, both proximal and distant, generally require a "leap of faith," a belief that one's partner will keep one's own best interests at heart and will try to maintain the relationship (Murray & Holmes, 1997; Murray, Holmes, Griffin, Bellavia, & Rose, 2001). One must have faith that a partner is both motivated (i.e., in love and committed) and actively behaving in a manner to maintain the relationship (i.e., not seeking alternative partners) to reciprocate similar behaviors and build intimacy and stability (Murray, Holmes, & Collins, 2006). Individuals who have a high sense of relational security should be more likely to work actively to maintain the stability of the relationship and therefore less likely to break up. Individuals with low relational security, who have little faith in either their partner's regard for them or the viability of the relationship, should invest less in the relationship, behave in a less committed manner, and ultimately initiate or cause the demise of their relationship. Accordingly, we predicted that relational security would be associated with greater relational stability in both long- distance and same-city relationships (Hypothesis 1). Previous research findings support such an expectation. For example, MacDonald and Ross found that people with greater optimism about the future of their relationship were less likely to break up, and Kurdek (2002) found the same for people who reported greater trust in their partners. Le, Smoak, and Agnew's (2006) meta-analysis of research on the stability of dating relationships demonstrated that individuals who report high trust are much less likely to break up.
Both individual characteristics and environmental factors can influence relational security, which in turn affects relational stability. Of all of the individual differences that relationship researchers have investigated, NA appears to have the strongest influence on relational stability (Hendrick, Hendrick, & Adler, 1988; Karney & Bradbury, 1995). High-NA individuals seem to possess several tendencies that make their relationships less likely to succeed than the relationships between low-NA individuals. In contrast to low-NA individuals, those with high-NA tend to (a) experience more negative emotions, which increases the negativity of their interactions with others (Furr & Funder, 1998; Vittengl & Holt, 1998); (b) worry more about how their partners view them and consequently report lower levels of relational security (Murray, Holmes & Griffin, 2000); (c) react to stressful experiences with greater negative affect (Bolger & Zuckerman, 1995; Marco & Suls, 1993); and (d) experience greater motivation to protect the self (Baumeister, Tice, & Hutton, 1989).
The self-protective, low risk-taking tendencies of high-NA individuals (Baumeister et al., 1989) may lead them to preemptively end their relationships in times of stress or relational threat. Alternatively, research involving both daily diary studies and laboratory studies has demonstrated that high-NA or rejection- sensitive individuals tend to respond to threats to relationship security with more negative, harmful interpersonal behaviors, such as devaluing their partners and distancing themselves from them (Downey, Feldman, & Ayduk, 2000; Downey, Freitas, Michaelis, & Khouri, 1998; Murray, Bellavia, Rose, & Griffin, 2003; Murray, Holmes, MacDonald, & Ellsworth, 1998). By withdrawing from their partners, insecure individuals presumably shield themselves from possible rejection (Murray, 2005; Murray & Derrick, 2005). Nevertheless, these behaviors lead the partners of high-NA individuals to feel reduced relationship satisfaction, which encourages contemplation about ending the relationship (Downey et al., 1998; Pietrzak, Downey, & Ayduk, 2005). Thus, the relationships of high-NA individuals, when under duress, may be more likely to end because these individuals have behaved so poorly that their partners take action to terminate the relationship. Given these tendencies toward self-protection and retaliation rather than toward relationship maintenance, we expected that high NA would be associated with low relational stability (Hypothesis 2), as previous researchers have found (see Karney & Bradbury, 1995, for a meta- analysis).
High NA alone does not necessarily lead relationships to terminate prematurely. Rather, the reactions of high-NA individuals to situations that threaten the security of their relationship-for instance, conflict (Downey et al., 2000; Downey et al., 1998) and stress (Bolger & Zuckerman, 1995; Marco & Suls, 1993)-contribute to a relationship's demise. In the present study, we expected that distant dating would serve as one of those situations that elicits threats to relational security and ultimately reduces the stability of relationships for high NA individuals. Long-distance dating can disrupt relational security, making partners worry about the future of their relationship (Aylor, 2003). In comparison with same-city couples, individuals in long-distance relationships are less able to monitor their partner's behavior and offer emotional support (e.g., Guldner, 1996). Research indicates that the majority of undergraduate students believe that long-distance relationships are less stable than same-city relationships (Helgeson, 1994a) and that distance worsens existing relationships (Knox, Zusman, Daniels, & Brantley, 2002). This belief might further reduce the perceived relational security of long-distance couples. Indeed, researchers have found that the physical separation involved in long-distance dating is associated with lower relationship satisfaction (Van Horn et al., 1997) and stability (Helgeson, 1994a; Lydon et al., 1997). Thus, we predicted that long-distance relationships would terminate earlier than would same-city relationships (Hypothesis 3). Moderation and Mediation: The Role of NA, Relational Security, and Long-Distance Status
Long-distance relationships appear to be especially threatening to relational security. Because high-NA individuals respond poorly to relational insecurity (Downey et al., 2000; Downey et al., 1998; Murray et al., 2003; Murray et al., 1998), we expected that high NA would be more damaging to the stability of long-distance relationships than to that of same-city relationships. In high-NA individuals, situations that elicit interpersonal risks, such as long-distance dating, should trigger (a) greater suspicions about a partner's regard, (b) more negative emotions, and (c) greater engagement in self-protective behaviors. Thus, we predicted that high-NA individuals in long-distance relationships would be more likely to break up with their partner within 1 year than would both high-NA individuals in same-city relationships and low-NA individuals in both types of relationships. However, we reasoned that gender, a variable known to impact multiple aspects of heterosexual relationships (see Peplau & Gordon, 1985, for a review), would be essential for predicting the impact of NA on relational stability in long-distance and same-city relationships.
We obtained assessments from both dating partners about their relationship, which enabled us to examine whether the NA levels of men and women differentially predicted the dissolution of long- distance relationships. The current literature on gender differences in relationship research is extensive, but findings are mixed. In a prominent review of the relationship literature, Karney and Bradbury (1995) found that individual characteristics of both husbands and wives predicted marital stability to about the same extent. In an earlier review of the literature, however, Barry (1970) reported that assessments obtained from husbands were generally stronger predictors of marital stability than were comparable assessments obtained from wives. More recent research (e.g., Kinnunen & Pulkkinen, 2003) has suggested that some individual differences predict stability for men, whereas other traits predict stability for women.
Differences in the predictability of men's and women's assessments may be particularly evident in long-distance dating relationships. Compared with women, men (a) place greater importance on face-to-face contact (Carpenter & Knox, 1986), (b) are less likely to build intimacy through distant communication methods (e.g., e-mail; Boneva, Kraut, & Frohlich, 2001), and (c) are less satisfied with and less able to adjust to long-distance relationships (Dellmann-Jenkins, Bernard-Paolucci, & Rushing, 1994; Helgeson, 1994b). Research on married couples indicates that for social support, men generally rely on their wives, whereas women depend heavily on friends and family (Antonucci & Aikyama, 1995; Fischer & Phillips, 1982; van Daalen, Sanders, & Willemsen, 2005). Long-distance dating may especially affect the relational security of men because they have reduced contact with their primary source of social support. If long-distance dating is more threatening to men's relational security, then NA may be more predictive of breakup in long-distance relationships when the partner with high NA is a man rather than a woman. In other words, we added greater specification to the prediction that high NA would be more detrimental in long-distance relationships than in same-city relationships. Specifically, we predicted that long-distance dating relationships in which the male partner has high NA would be more likely to have dissolved within 1 year of the study than would either (a) distant relationships with low-NA men or (b) same-city relationships regardless of the NA status of the men (Hypothesis 4). We also predicted that in long-distance relationships, men's NA would be more predictive of breakup than would women's NA (Hypothesis 5).
Long-distance dating should be a particularly stressful circumstance for high-NA males, one that ultimately reduces their relational security and leads to the early end of these relationships. Men's greater value of face-to-face contact and their primary reliance on female partners for social support (Antonucci & Aikyama, 1995; Fischer & Phillips, 1982; Van Daalen, Sanders, & Willemsen, 2005) may lead high-NA men to have reduced levels of trust and faith in a partner. Decreasing social support and contact with a partner is associated with reductions in relational security (Parks & Adelman, 1983) and stability (Baxter, 1986). Thus, the reduced relational security experienced by high-NA men in distant relationships likely leads these men to either initiate a breakup of these relationships or behave so poorly toward their partners that their partners end the relationships. We therefore anticipated that men's ratings of relational security would mediate the relation between men's NA and the dissolution of long-distance relationships (Hypothesis 6).
We conducted the present study to investigate how NA, gender, and long-distance status interact to predict relational stability. To our knowledge, researchers have not previously integrated research on NA, gender, and relationship security into predicting the stability of long-distance relationships. Several scholars, however, have called on researchers to investigate such factors. For example, in recent reviews of the literature, Rhodes (2002) called for more researchers to investigate personality, and Aylor (2003) suggested that more researchers investigate trust-a central component of relational security-in distant relationships. We believe that by investigating these variables in the context of long-distance relationships, we can identify which long-distance relationships are more likely to remain intact and which are likely to dissolve.
We designed this study to test the aforementioned six hypotheses. The first three hypotheses relate to previous research whose results we hoped to replicate, whereas the latter three hypotheses are previously untested in the literature. To test our predictions, we conducted a 1-year longitudinal study with same-city and long- distance dating couples. At the outset (Time 1), both partners completed questionnaires assessing different components of NA and relational security. Researchers contacted participants 2 months, 6 months, and 1 year later to determine whether the couples had remained together.
Participants were 91 introductory psychology students who volunteered to participate with their current dating partner. To qualify for the study, potential participants had to be in an exclusive dating relationship that had started more than 3 months earlier. Sixty-eight couples completed the questionnaires in the laboratory. For the remaining 23 couples, one partner completed the survey in the laboratory, and the other partner, who was unable to attend, completed the survey at home and returned it by mail. Individuals recruited from the introductory psychology class received partial course credit; their partners not in the class received $10.
Participating couples reported dating exclusively for an average of 12.56 months (SD = 8.76 months, range = 3-60 months). The average age of participants was 19.45 years (SD = 1.56 years, range = 16-27 years). Participants also indicated whether they considered their relationship to be long distance. We verified their self-described status by comparing the area codes of partners' phone numbers. In all cases, self-described status matched the status derived from area-code comparisons. Thirty-seven (42%) of the couples were in long-distance relationships. Participants in same-city and long- distance relationships did not differ in age or length of relationship, Fs(1, 180)
Participants completed the survey after we had obtained their informed consent. Participants who completed the survey in the laboratory were separated by partitions and instructed not to communicate with one another as they completed the questionnaires in groups of up to 6 individuals. Participants who submitted their surveys through the mail were instructed not to discuss the survey with their partner prior to returning it. To assess relational stability-our dependent variable in all six hypotheses-we telephoned participants 2 months, 6 months, and 1 year after the initial survey. If we did not reach participants after three phone calls on 3 different days at 3 different times, we contacted them at another phone number (e.g., their parents' phone number) or by e-mail. Participants indicated whether they were still with their original partner and, if the relationship had ended, when the breakup had occurred. At least one member from each couple participated in this short interview at each time period. We were able to obtain information on relationship status 1 year after the initial survey for 100% of the couples. After the 1-year period, we provided all participants with feedback on the original research goals and the general findings from the study. Measures of NA at Time 1
We assessed NA with three standard scales that measure (a) self- esteem, (b) depression, and (c) pessimism.
Self-esteem. To assess self-esteem, we used Rosenberg's (1965) Self-Esteem Scale (RSES; as = .88 and .87 for men and women, respectively, in the present study). Participants responded to each of the 10 items on the RSES (sample item: "I feel that I am a person of worth, at least on an equal basis with others") on a 9-point scale ranging from 1 (very strongly disagree) to 9 (very strongly agree). After reverse coding the negatively worded items, we averaged each individual's scores on the 10 items to create a total score whereby high scores indicated high self-esteem.
Depression. To assess depression, we asked participants to complete Beck's (1967) Depression Inventory (BDI; as = .85 for both men and women in the present study). The BDI contains 21 groups of four related statements. Respondents select one of the four statements in each grouping that best describes their current emotional and physical state; responses are then coded from 0 to 3 to represent increasing degrees of symptom severity. For example, the first group of statements was as follows: (a) "I do not feel sad" (coded as 0); (b) "I feel sad" (coded as 1); (c) "I am sad all the time and I can't snap out of it" (coded as 2); and (d) "I am so sad or unhappy that I can't stand it" (coded as 3). Scores are then summed to indicate presence and severity of depression; they can range from 0 (no depression) to 63 (high depression ).
Pessimism. Participants responded to three items from the revised Life Orientation Test (LOT-R; Scheier, Carver, & Bridges, 1994) designed to measure dispositional pessimism (as = .73 and .63 for men and women, respectively, in the present study). Participants responded to each of the three items (sample item: "If something can go wrong for me, it will") on a 5-point scale ranging from 1 (disagree a lot) to 5 (agree a lot). We then averaged scores to create a total pessimism score whereby high values indicated greater pessimism.
To create a composite score of negative affectivity for each participant, we first reverse scored the self-esteem average so that for all three scales, lower numbers reflected lower levels of negative affectivity (i.e., higher self-esteem, lower depression, and lower pessimism). We then standardized and combined scores on the three measures to create a reliable index of negative affectivity (as = .87 and .84 for men and women, respectively).
We selected these scales because in addition to demonstrated reliability in the present study, all have demonstrated high performance in similar samples. The RSES is the most widely used of the various self-esteem measures (Blascovich & Tomaka, 1991; Gray- Little, Williams, & Hancock, 1997); it has also received the most empirical validation (Byrne, 1996; Gray-Little et al.; Wylie, 1989), and Wylie recommended its use across a variety of populations, including university and college samples. The BDI is the most widely used self-report-based depression measure in empirical studies (Shaver & Brennan, 1991) and is appropriate for use in nonclinical college samples (Kendall, Hollon, Beck, Hammen, & Ingram, 1987). Last, the LOT-R is the most widely accepted measure of pessimism and optimism (see Chang, 2001, for examples) and was originally validated on college samples (Scheier et al., 1994).
Measures of Relational Security at Time 1
We measured three variables-faith in partner, expected support, and relationship optimism-to assess relational security. Unless otherwise noted, participants responded to items on the measures on a 9-point scale ranging from 1 (not at all true) to 9 (extremely true). Some items were reverse coded so that for all three measures, higher scores would represent greater security. We then averaged each participant's scores on each measure.
Faith in partner. Participants responded to the following four items from the Faith subscale of Rempel, Holmes, and Zanna's (1985) Trust Scale: (a) "I feel completely confident that my partner loves me"; (b) "I feel that I can trust my partner completely"; (c) "Though times may change and the future is uncertain, I know my partner will always be ready to offer me strength and support"; and (d) "I am confident that my partner will always be able to see the best in me, no matter how poorly I behave" (as = .75 and .73 for men and women, respectively, in the present study).
Expected support. Participants indicated how likely it was that their partner would engage in four different support-related behaviors (derived from Murray et al., 1998) in the next 6 months. Participants responded to the following items: (a) "My partner will want more independence and pull away from me" (reverse coded); (b) "My partner will neglect my needs and feelings" (reverse coded); (c) "My partner will forgive me if I disappoint him or her"; and (d) "My partner will provide needed support when I'm feeling down" (as = .62 and .61 for men and women, respectively, in the present study).
Relationship optimism. Participants indicated how confident they were that they and their partner (a) would be together 5 years from now, (b) would get married, and (c) would remain together for life (derived from MacDonald & Ross, 1999). A response of 100% to each item indicated that participants were completely certain this outcome would occur, a score of 50% indicated that they were completely uncertain about that outcome, and a score of 0% indicated that they were completely certain that outcome would not occur (as = .98 for both men and women in the present study).
We standardized each participant's mean score for faith in partner, relationship optimism, and expected support and collapsed across them to create an index of relationship security (as = .86 and .76 for men and women, respectively, in the present study).
Our first task was to determine whether any of the couples had terminated their relationships within the assessment periods; because all six of our hypotheses involved relational stability, any further analysis would have been futile if all couples had remained intact. However, this was not the case with the present sample: Four of the couples (4.4%) had separated at 2 months, 11 (12.1%) had separated at 6 months, and 26 (28.6%) had separated at 1 year. Because of low variability in breakup at the 2- and 6-month assessments, we conducted analyses only on the breakup rate at the 1- year period. In the main analyses to test the six hypotheses, we adjusted for the interdependence of the data. Because we assessed both members of the couple, the reports of men and women described the same relationship and thus had a high probability of being interrelated. Consequently, the dyad was the unit of analysis. That is, men's and women's NA and relational security represented four variables within the data file: (a) men's NA, (b) women's NA, (c) men's relational security, and (d) women's relational security. We treated each of these predictors as a within-subject variable. Two variables, however, were independent of the other cases: proximity status and relationship outcome. Because there was only one status and one outcome per couple, we treated these variables as between- subject variables. We discuss below the strategy used to adjust for the interdependence of the data.
Predicting Relational Stability
To test the first and second hypotheses, we examined whether measures of relational security and NA obtained at Time 1 predicted relationship breakup. We conducted zero-order point-biserial correlations between each of the Time 1 indexes and relationship status at 1 year (coded as 0 for broke up and 1 for intact) for the entire sample. Overall, relational security was positively associated with relational stability for men, r(90) = .37, p
Next, we tested whether the proximity of partners predicted subsequent breakup (Hypothesis 3). A cross-tabulation of proximity status (long distance vs. same city) and stability outcome (together vs. broken up) revealed that a nonsignificantly greater proportion of long distance relationships had dissolved (35% vs. 24%), .2(1, N = 91) = 1.32, ns. Thus, our first three predictions were confirmed, and our results replicated the previous findings on which our research was based. We next tested the previously untested hypotheses.
To test Hypothesis 4-that men's NA in long-distance relationships would be more predictive of relationship outcomes than would either (a) men's NA in same-city relationships or (b) women's NA regardless of proximity-we conducted a hierarchical multiple regression analysis with relationship status at 1 year (coded as 0 for apart and 1 for intact) as the criterion variable and the dyad as the unit of analysis (see Kashy & Kenny, 2000). Per Aiken and West's (1991) recommendation, we centered the continuous variables of men's and women's relationship security and NA by subtracting the mean for each variable from each score to yield a mean score of 0. We effect- coded proximity (long distance = -1, same city = 1). In Step 1 of the analysis, we entered as predictors (a) men's relationship and NA indexes, (b) women's relationship and NA indexes, and (c) the long- distance status of the relationship (see Table 1 for intercorrelations between variables). Step 1 tested for any main effects that these variables may have had on relationship outcome. In Step 2, we entered simultaneously as predictors the two-way interactions (a) between men's and women's relational security and NA indexes and (b) between these indexes and long-distance status. In Step 3, we entered the three-way interactions (a) between men's NA, relational security, and proximity and (b) between women's NA, relational security, and proximity.2 Step 1 (including main effects), Step 2 (including the two-way interactions), and Step 3 (including three-way interactions) were all significant, R = .42, F(5, 85) = 3.63, p
As Table 2 shows, the more favorably that men evaluated the security of their relationship, the more likely were both same-city and different-city couples to be intact 1 year later. A significant interaction between men's levels of NA and proximity revealed that men's NA was significantly related to relational stability only in long-distance couples (see Figure 1). For long-distance relationships, the greater was the men's NA, the less likely were these relationships to be intact 1 year later, B = -0.32 (SE = 0.09), ss = -.53, t(36) = -3.67, p
Testing gender differences. We sought to test whether men's NA predicted stability in long-distance relationships more than did (a) men's NA in same-city relationships and (b) women's NA in both types of relationships. Because we had collected data from both members of the couple, we could not run the straightforward test of the predicted three-way interaction between gender, NA, and proximity (Hypothesis 5). Instead, we had to compare the significance of the two-way interaction between men's NA and proximity to the interaction between women's NA and proximity. A finding that the interaction between men's NA and proximity was significantly more predictive of relationship outcome than was the interaction between women's NA and proximity would have supported Hypothesis 5 (that it is men's and not women's NA that predicts relationship outcomes in long-distance relationships).
To examine whether the men's NA x Proximity interaction was significantly more predictive than was the women's NA x Proximity interaction, we created and compared two structural equation models using Amos 4.0 (Arbuckle & Wothke, 1999). In both models, we included (a) men's NA, (b) women's NA, (c) men's relational security, (d) women's relational security, and (e) proximity as predictors of relational stability (see Table 2) and allowed all predictors to covary with each other. The first model represented a situation in which no gender differences existed (no gender difference model). We constrained paths for each set of predictors (NA, relational security, NA x Proximity, and Relational Security x Proximity) so they would be equal for men and women. For example, we set men's NA and women's NA to equally predict relational stability. This model fit the data very well, .2(5) = 5.72, ns, CFI = .997, RMSEA = .040. In the second model, we constrained all paths so they would be equal across genders, except for the NA x Proximity interactions (predicted gender difference model). This model fit the data extremely well, .2(4) = 2.88, ns, CFI = 1.00, RMSEA = .000. We conducted a chi-square difference test to compare the no gender difference model to the predicted gender difference model. Results indicated that the predicted gender difference model, in which the NA x Proximity interactions were not constrained, fit the data marginally better than did the no gender difference model, .2(1) = 2.84, p = .09. Thus, the interaction between men's NA and proximity was more predictive of relational stability than was the interaction between women's NA and proximity, supporting Hypothesis 5.
Mediation analysis. We conducted a test of mediated moderation, as outlined by Baron and Kenny (1986), to examine whether the effect of the interaction between men's NA and proximity on relational stability was mediated by men's perceptions of relational security (Hypothesis 6). In the first step, in which we controlled for lower- order main effects, men's NA x Proximity was significantly associated with relational stability, B = 0.16 (SE = 0.08), ss = .29, t(87) = 2.84, p
Testing Alternative Explanations
In theory, differences between men and women on the initial measures of NA and relationship security could have impacted our findings on predictions of relational stability. To rule out this possible explanation for the gender differences we found in predicting stability, we tested for gender differences in our initial measures by conducting a 2 (gender: men's measure vs. women's mea-sure) x 2 (proximity: long distance vs. same city) repeated-measures analysis of variance (ANOVA) for each of the components of NA (self-esteem, depression, pessimism), with the dyad as the unit of analysis. In this analysis, gender was a within- subject factor-because the responses of the woman and the man in a couple were potentially interrelated-whereas proximity was a between- subject factor. We could not directly run this test on the index of NA because each of the measures of NA used a different scaling system, and thus the systems were standardized to form the index of NA. Therefore, any test of mean gender differences on NA would ultimately lead to nonsignificance because the means of men's NA and women's NA (being two separate variables) would always be zero on the standardized index. There were no significant main effects and no significant interaction for any of the NA components. Thus, gender differences in initial levels of NA could not have accounted for the greater impact that high NA in men had on the stability of long-distance relationships (see Table 3 for means, standard deviations, and correlations between partner's scores on these measures).
We next examined whether possible gender differences in relational security could have contributed to men's NA being a stronger predictor of stability for long-distance relationships. To test this alternative explanation, we conducted a 2 (gender: men's measure vs. women's measure) x 2 (proximity: long distance vs. same city) repeated-measures ANOVA. Again, gender was a within- subject factor and proximity was a between-subject factor. As in the above analyses, we ran this test separately on each component of relational security. There were no significant main effects for proximity and no significant interactions between gender and proximity on any of the relational security measures. There were, however, unexpected main effects for gender on the relational security scales: Compared to their male partners, women (a) were more optimistic about the future of their relationship, F(1, 89) = 6.86, p
As long-distance relationships become increasingly more common (Guldner, 2003), it becomes essential that relationship researchers focus on the dynamics and processes of these relationships (see Aylor, 2003; Stafford, 2004, 2005). Although we do not propose that the processes in these relationships are qualitatively different from those in geographically close relationships, we believe that being geographically distant from a partner enhances interpersonal risk and stress, and certain relationship processes may therefore be exaggerated in long-distance relationships. Indeed, our results largely confirmed these expectations. Our findings support our first three hypotheses, which relate to replicating previous work: Lower relational security and high NA were associated with reduced relational stability (Hypotheses 1 and 2), and long-distance relationships were somewhat less likely to endure than were same- city relationships (Hypothesis 3). More important, two of our previously untested hypotheses were confirmed: Men's NA was significantly associated with relational stability in long-distance but not same-city relationships (Hypothesis 4), and men's NA in long- distance relationships significantly predicted stability, whereas women's NA did not (Hypothesis 5). However, our expectation regarding the mechanism of this effect was not confirmed: Relational security did not mediate the impact of men's NA on stability in long- distance relationships (Hypothesis 6). In contrast to the majority of researchers who have investigated the predictors of relational stability in dating couples, we assessed relational security and the individual characteristics of both partners. Also, we included dating couples experiencing a particularly interpersonally risky situation-living in different cities-as well as couples living in the same city. Past research findings have been inconsistent with respect to whether the assessments of men and women are differentially predictive of relational stability (Barry, 1970; Karney & Bradbury, 1995). Our findings provide support for both Barry's and Karney and Bradbury's conclusions. When examining the entire sample, we found that the pattern of correlations was very similar for men and women, as Karney and Bradbury's review would suggest. When we included the proximity variable, however, the results indicated a different relation. The multiple regression analyses indicate that the association between stability and men's NA hovers near zero in same-city relationships but becomes statistically significant in long-distance relationships. Moreover, as revealed in the structural equation analysis, the interaction between NA and proximity is significantly more predictive for men than for women. In essence, men with high NA seem to contribute to the dissolution of long-distance but not same-city relationships to a greater degree than do high-NA women.
One possible explanation for the lack of association between women's NA and relational stability is that the women's NA scores may have a truncated range (i.e., reduced variance) or may be generally higher than are men's. However, we tested for and did not find any gender differences in means, variances, or distributions on any of the NA variables. Another possibility is women's assessments may be less reliable than men's, again reducing the possibility of finding significant associations between variables for women. Importantly, however, women's NA was not less reliable than men's NA.
Although the zero-order correlations and regression analyses yielded largely consistent findings, there is one important difference: Both men's and women's assessments of relational security were associated with relational stability in the zero- order correlations, but only men's assessments predicted stability in the hierarchical regression. We likely did not obtain a prediction for women's assessments in the regression analysis because partners' evaluations of relational security were moderately related, r(90) = .49, p
Unexpectedly, women's reports of relational security were higher than those of their male partners. We examined whether selection factors may account for our finding that women were more positive about the future of their relationships at Time 1 than were men. Because we recruited couples from an introductory psychology course in which approximately 70% of the students were women, the majority of participants recruited from the course were women (n = 60). Conceivably, individuals who volunteer to participate in relationship studies are more secure with their relationships. If so, it may explain why women, who mostly volunteered from the psychology course, reported higher relational security than did men, who may have been forced into the study by their partners. To examine this possibility, we compared reports of relational security from the men recruited from the psychology course to those from the men recruited by their partners; we then studied the same comparison for women. We found no statistical differences between men recruited from the psychology course and men recruited by their partners.
Strengths and Limitations
The majority of researchers studying dating couples have investigated the reports of only one partner. The use of such one- sided reports may have reduced the likelihood of finding significant gender differences. Our assessing reports from both partners may represent a more sensitive test of gender differences within relationships. In comparison to many previous researchers, we followed dating couples for a relatively long period of time after the initial survey. Although 1 year is short in comparison to the intervals examined in studies of marital stability, it is longer than the few months typically studied by researchers of dating relationships. Had we stopped at 2 or even 6 months, we may have missed the predicted interaction between men's NA and long-distance dating because the variation in relational stability may have been too low.
We also measured and tested psychological processes that could explain why high NA in men is associated with reductions in relationship longevity in long-distance couples. High-NA men's experiences of diminished faith in their partners' regard and support may be exacerbated by the physical separation from a partner. Relative to their high-NA female counterparts who have substitute support networks, high-NA men may experience a greater reduction in social support and approval (Kernis, Cornell, Sun, Berry, & Harlow, 1993; Leary, Tambor, Terdal, & Downs, 1995) in long- distance relationships. Unexpectedly, on average, men in long- distance relationships did not experience less security in their relationships than did men in same-city relationships, and relational security did not mediate the relationship between the men's NA x Proximity interaction and relational stability. In retrospect, we should have assessed relational security at each follow-up session because a reduction in relational security of high- NA men may have occurred after longer periods of physical separation. However, another possible explanation for why Hypothesis 6 was not supported is that the measures of relational security, which have received less psychometric investigation than have the measures of NA, are not sufficiently reliable. Although the internal consistencies of these scales are comparable to those for NA in this sample, the validity of these scales is not as well established (see Fletcher, Simpson, & Thomas, 2000).
Alternatively, other psychological processes could account for why high NA in men is detrimental to their long-distance relationships. For instance, previous researchers have demonstrated that individuals with high NA find life events more stressful than do low-NA individuals (Bolger & Zuckerman, 1995; Marco & Suls, 1993). Diminished support from a partner may lead high-NA men to experience greater distress rather than reduced relational security, which may in turn diminish the likelihood of a relationship's success. Thus, at each follow-up assessment, we should have measured not only relational security but also distress. Future researchers should include such aspects in any further investigation of these effects.
One strength of our research is that we were able to collect both longitudinal data from both partners and stability data from 100% of the original couples. We attribute this success to our extremely short follow-up interviews. However, this success is paired with the drawback that we did not assess relational security or NA at these follow-up assessments.
Although our study had many strengths-including our (a) confirming most hypotheses, (b) investigating a novel situation, (c) exploring the mechanism behind a predicted effect while including both members of the couple, and (d) obtaining outcome data from all couples in the sample-it also had several limitations. As previously mentioned, in retrospect we should have included measures of relational security, NA, and distress at each of our follow-up assessments to fully explore change and the possible mechanism behind men's NA being a more significant predictor in long-distance relationships. Three factors that we neglected to assess may be critical to the success of long-distance dating relation-ships: (a) actual geographical distance, (b) frequency and amount of contact, and (c) periods of geographical closeness. Although actual geographic distance may be an important fact in the stability of long-distance relationships, we chose to define a long-distance relationship on the basis of whether the partners identified themselves as in such a relationship, which is similar to the method used by Dellmann-Jenkins et al. (1994) and recommended by Stafford (2004) in a recent review of the literature. Because of this decision, we neglected to assess the mileage between residences, which Carpenter and Knox (1986) deemed crucial to understanding long- distance relationships. Regardless of our definition of long- distance dating, by not assessing actual distance between the partners, we may have missed an important predictor.
Although some researchers have reported that the actual geographical distance between partners may be important (e.g., Carpenter & Knox, 1986), certainly the amount and type of contact between long-distance partners is essential to the success of long- distance relationships. Several researchers have reported that greater face-to-face contact is associated with greater satisfaction and trust (Dainton & Aylor, 2002; Holt & Stone, 1988) and with relational stability (Carpenter & Knox). The frequency of letters (Stafford & Reske, 1990), telephone calls, and e-mails (Dainton & Aylor) between partners is also associated with increased satisfaction and relational security. However, other researchers have argued that contact may actually have potentially negative effects for long-distance relationships, such as increased conflict (Sahlstein, 2004) and disillusionment (Gerstel & Gross, 1984). In the current study, we did not assess the amount or type of contact and therefore cannot address whether these variables moderated or mediated any of the results reported here. Our long-distance dating sample consisted of first-year university students, a population that may experience periods of frequent geographical closeness followed by periods of distance. Many of the relationships in the present sample became long distance when one or both partners moved away to attend university or college. It is not uncommon for university students who move away to attend university during the fall and winter to return to their primary residence during the summer. If both members of a couple were to do this, they could essentially spend the summers as same-city couples. Unfortunately, we did not assess whether couples in the present study experienced periods of geographical closeness, despite Stafford's (2005) recent appeal for researchers to investigate such circumstances.
Directions for Future Research
Despite the strengths of our approach, we were unable to discover the mechanism that explains why men's NA is more predictive of the breakup of long-distance couples than is women's NA. To track underlying psychological processes more precisely, future researchers could include a more diary-like accounting procedure, in which individuals in long-distance relationships report relationship- relevant episodes and their feelings and assessments on a regular basis. At the very least, researchers should investigate participants' relational security, NA, and distress on a regular basis. Presumably, declines in relational security and increases in distress and NA predict breakup in long-distance couples, and couples with high-NA men are more likely to experience a rapid decline in these areas. Researchers may find ongoing assessments of the psychological and physical well-being of couples in long- distance relationships to be particularly valuable. If high-NA men experience less social support, they may also be more vulnerable to the negative health consequences accompanying a reduction in social support (e.g., Reis, 1984) and the dissolution of a relationship (e.g., Bloom, Asher, & White, 1978).
Ongoing assessments may also assist researchers in understanding how the amount and type of contact and periods of geographical closeness may mediate or moderate the impact of men's NA on relational stability in long-distance couples. High-NA men should experience particular difficulty when prevented from having regular face-to-face contact with their partners (Carpenter & Knox, 1986). These individuals may also experience distress when telephone and e- mail contact is unavailable. A vicious cycle may exist in these contexts: When contact with a distant partner becomes less frequent, high-NA men may experience reduced relational security, which may lead them to distance themselves emotionally from their partners and may reduce their motivation to initiate contact with a distant partner. Such behaviors would surely cause concern for the female partner and may discourage her from initiating contact. Relationship breakup may be an inevitable outcome in such circumstances. Alternatively, high-NA men may be more vulnerable to the increased conflict (Sahlstein, 2004) and disillusionment (Gerstel & Gross, 1984) to which long-distance couples are susceptible when they become geographically close or when the amount of contact is high. Again, a daily-diary approach would be most useful for future investigations of this possibility.
Conclusions and Implications
Our results indicate that not all couples need worry about the viability of long-distance dating relationships. Couples with high- NA men, however, should be wary of maintaining a long-distance relationship and perhaps should take steps to find other sources of social support or to boost their levels of security. Mental health professionals should be especially aware of the possibility that the high NA males in long-distance relationships in their practice may be at increased risk of breakup and the negative consequences of such relationship dissolution.
One of the most important contributions of this study is that our findings emphasize that individual characteristics do not exist in a vacuum separated from context. The interaction between situational context (e.g., living apart) and enduring vulnerabilities (e.g., high NA) best predicts relationship outcomes. Consistent with an interactionist perspective (Endler & Parker, 1992), our results highlight the importance of investigating the interaction between personal characteristics and situations in relationships. In the present study, the interaction occurred exclusively for men, but this finding is likely a product of the particular context that we investigated. If we had studied a context that is more likely to heighten insecurity in women than in their male partners (e.g., transition to parenthood; see Osofsky, 1985; Ross & Van Willigen, 1996), we may have found that women's NA is more predictive of relational stability than is men's NA.
1. Correlation analyses using the individual measures instead of the indexes replicated a similar pattern. To conserve space, we report here results only for the indexes. We performed subsequent analyses only with the indexes.
2. We selected only a few predicted and possible interaction terms to include in the presented regression analysis. However, to ensure that we had not missed a statistically important interaction, we conducted a hierarchical linear regression with the terms included in the presented regression (Table 2) in Step 1 and all other possible interaction terms in Step 2. Step 1 was significant, F(11, 79) = 3.16, p
3. We replicated these analyses using logistic regression. The results are virtually identical to those presented here.
Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions. Newbury Park, CA: Sage.
Antonucci, T., & Aikyama, A. (1995). Convoys of social relationships: Family and friendships within a life span context. In R. Blieszner & V. H. Bedford (Eds.), Handbook of aging and the family (4th ed., pp. 355-371). Westport, CT: Greenwood Press.
Arbuckle, J. L., & Wothke, W. (1999). AMOS 4.0 user's guide. Chicago: SmallWaters.
Aylor, B. A. (2003). Maintaining long-distance relationships. In D. J. Canary & M. Dainton (Eds.), Maintaining relationships through communication: Relational, contextual, and cultural variations (pp. 127-139). Mahawh, NJ: Erlbaum.
Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173-1182.
Barry, W. A. (1970). Marriage research and conflict: An integrative review. Psychological Bulletin, 73, 41-54.
Baumeister, R. F., Tice, D. M., & Hutton, D. G. (1989). Self- presentational motivations and personality differences in self- esteem. Journal of Personality, 57, 547-579.
Baxter, L. A. (1986). Gender differences in the heterosexual relationship rules embedded in breakup accounts. Journal of Social and Personal Relationships, 3, 289-306.
Beck, A. T. (1967). Depression: Causes and treatment. Philadelphia: University of Pennsylvania Press.
Blascovich, J., & Tomaka, J. (1991). The self-esteem scale. In J. P. Robinson, P. R. Shaver, & L. S. Wrightsman (Eds.), Measures of personality and social psychological attitudes (pp. 115-160). San Diego, CA: Academic Press.
Bloom, B. L., Asher, S. J., & White, S. W. (1978). Marital disruption as a stressor: A review and analysis. Psychological Bulletin, 85, 867-894.
Bolger, N., & Zuckerman, A. (1995). A framework for studying personality in the stress process. Journal of Personality and Social Psychology, 69, 890-902.
Boneva, B., Kraut, R., & Frohlich, D. (2001). Using e-mail for personal relationships: The difference gender makes. American Behavioral Scientist, 45, 530-549.
Byrne, B. M. (1996). Measuring self-esteem across the lifespan: Issues and instrumentation. Washington, DC: American Psychological Association.
Carpenter, D., & Knox, D. (1986). Relationship maintenance of college students separated during courtship. College Student Journal, 20, 86-88.
Cate, R. M., Levin, L. A., & Richmond, L. S. (2002). Premarital relationship stability: A review of recent research. Journal of Social and Personal Relationships, 19, 261-284.
Chang, E. C. (2001). Optimism and pessimism: Implications for theory, research, and practice. Washington, DC: American Psychological Association.
Dainton, M. & Aylor, B. (2002). Patterns of communication channel use in the maintenance of long-distance relationships. Communication Research Reports, 19, 118-129.
Dellmann-Jenkins, M., Bernard-Paolucci, T. S., & Rushing, B. (1994). Does distance make the heart grow fonder? A comparison of college students in long distance and geographically close dating relationships. College Student Journal, 28, 212-219.
Downey, G., Feldman, S., & Ayduk, O. (2000). Rejection sensitivity and male violence in romantic relationships. Personal Relationships, 7, 45-61.
Downey, G., Freitas, A. L., Michaelis, B., & Khouri, H. (1998). The self-fulfilling prophecy in close relationships: Rejection sensitivity and rejection by romantic partners. Journal of Personality and Social Psychology, 75, 545-560. Endler, N. S., & Parker, J. D. (1992). Interactionism revisited: Reflections on the continuing crisis in the personality area. European Journal of Personality, 6, 177-198.
Fischer, C. S., & Phillips, S. L. (1982). Who is alone? Social characteristics of people with small networks. In L. A. Peplau & D. Perlman (Eds.), Loneliness: A sourcebook of current theory, research, and therapy (pp. 21-39). New York: Wiley.
Fletcher, G. J. O., Simpson, J. A., & Thomas, G. (2000). The measurement of perceived relationship quality: A confirmatory factor analysis approach. Personality and Social Psychology Bulletin, 26, 340-354.
Furr, R. M., & Funder, D. C. (1998). A multimodal analysis of personal negativity. Journal of Personality and Social Psychology, 74, 1580-1591.
Gerstel, N., & Gross, H. (1984). Commuter marriage: A study of work and family. New York: Guilford Press.
Gray-Little, B., Williams, V. S. L., & Hancock, T. D. (1997). An item response theory analysis of the Rosenberg Self-Esteem Scale. Personality and Social Psychology Bulletin, 23, 443-451.
Guldner, G. T. (1996). Long-distance romantic relationships: Prevalence and separation-related symptoms in college students. Journal of College Student Development, 37, 289-296.
Guldner, G. T. (2003). Long distance relationships: The complete g uide. Corona, CA: JF Milne.
Helgeson, V. S. (1994a). The effects of self-beliefs and relationship beliefs on adjustment to a relationship stressor. Personal Relationships, 3, 241-258.
Helgeson, V. S. (1994b). Long-distance romantic relationships: Sex differences in adjustment and breakup. Personality and Social Psychology Bulletin, 20, 254-265.
Hendrick, S. S., Hendrick, C., & Adler, N. L. (1988). Romantic relationships: Love, satisfaction, and staying together. Journal of Personality and Social Psychology, 54, 980-988.
Holt, P. A., & Stone, G. L. (1988). Needs, coping strategies, and coping outcomes associated with long-distance relationships. Journal of College Student Development, 29, 136-141.
Karney, B. R., & Bradbury, T. N. (1995). The longitudinal course of marital quality and stability: A review of theory, method, and research. Psychological Bulletin, 118, 3-34.
Kashy, D. A., & Kenny, D. A. (2000). The analysis of data from dyads and groups. In H. T. Reis & C. M. Judd (Eds.), Handbook of research methods in social and personality psychology (pp. 451- 478). Cambridge, England: Cambridge University Press.
Kendall, P. C., Hollon, S. D., Beck, A. T., Hammen, C. L., & Ingram, R. E. (1987). Issues and recommendations regarding the use of the Beck Depression Inventory. Cognitive Therapy and Research, 11, 289-299.
Kernis, M. H., Cornell, D. P., Sun, C. R., Berry, A., & Harlow, T. (1993). There's more to self-esteem than whether it is high or low: The importance of stability of self-esteem. Journal of Personality and Social Psychology, 65, 1190-1204.
Kinnunen, U., & Pulkkinen, L. (2003). Childhood socioemotional characteristics as antecedents of marital stability and quality. European Psychologist, 8, 223-237.
Knox, D., Zusman, M. E., Daniels, V., & Brantley, A. (2002). Absence makes the heart grow fonder? Long distance dating relationships among college students. College Student Journal, 36, 364-367.
Kurdek, L. A. (2002). Predicting the timing of separation and marital satisfaction: An eight-year prospective longitudinal study. Journal of Marriage and the Family, 63, 163-179.
Le, B., Smoak, N. D., & Agnew, C. R. (2006, July). A meta- analytic examination of predictors of dissolution in dating relationships. Poster presented at the conference of the International Association for Relationship Research, Rethymno, Crete.
Leary, M. R., Tambor, E. S., Terdal, S. K., & Downs, D. L. (1995). Self-esteem as an interpersonal monitor: The sociometer hypothesis. Journal of Personality and Social Psychology, 68, 518- 530.
Lydon, J., Pierce, T., & O'Regan, S. (1997). Coping with moral commitment to long-distance dating relationships. Journal of Personality and Social Psychology, 73, 104-113.
MacDonald, T. K., & Ross, M. (1999). Assessing the accuracy of predictions about dating relationships: How and why do lovers' predictions differ from those made by observers? Personality and Social Psychology Bulletin, 25, 1417-1429.
Marco, C. A., & Suls, J. (1993). Daily stress and the trajectory of mood: Spillover, response assimilation, contrast, and chronic negative affectivity. Journal of Personality and Social Psychology, 64, 1053-1063.
Murray, S. L. (2005). Regulating the risks of closeness: A relationship-specific sense of felt security. Current Directions in Psychological Science, 14, 74-78.
Murray, S. L., Bellavia, G. M., Rose, P., & Griffin, D. W. (2003). Once hurt, twice hurtful: How perceived regard regulates daily marital interactions. Journal of Personality and Social Psychology, 84, 126-147.
Murray, S. L., & Derrick, J. (2005). A relationship-specific sense of felt security: How perceived regard regulates relationship- enhancement processes. In M. Baldwin (Ed.), Interpersonal cognition (pp. 153-179). New York: Guilford Press.
Murray, S. L., & Holmes, J. G. (1997). A leap of faith? Positive illusions in romantic relationships. Personality and Social Psychology Bulletin, 23, 586-604.
Murray, S. L., Holmes, J. G., & Collins, N. L. (2006). Optimizing assurance: The risk regulating system. Psychological Bulletin, 132, 641-666.
Murray, S. L., Holmes, J. G., & Griffin, D. W. (2000). Self- esteem and the quest for felt security: How perceived regard regulates attachment processes. Journal of Personality and Social Psychology, 78, 478-498.
Murray, S. L., Holmes, J. G., Griffin, D. W., Bellavia, G., & Rose, P. (2001). The mismea-sure of love: How self-doubts contaminate relationship beliefs. P