Passion for life

Participants

Five hundred thirty-nine (41% men) registered voters in a medium-sized (population = 60,000) Midwestern community participated in a telephone survey about their current romantic relationship. Respondents ranged in age from 18 to 92 years (median (Mdn) = 49.3; 17 not reporting (nr)). Most (84.2%) respondents were living with their romantic partner (1 nr), and 74.7% were married to their partner. Of these, 14 (2.6%) were involved with a same-sex romantic partner (1 nr). Respondents’ relationships ranged in length from 1 day to 65 years (Mdn = 16.6 years; 38 nr).

Where possible, the sample demographics were compared with census data for the county in which the study was conducted (U.S. Census, 2000). These comparisons revealed that women (59% of the sample) were overrepresented compared to the local population (51% female). The median age of the sample (49 years) was slightly greater than for adults in the area (44 years), although the proportion of older adults (e.g., over 65 years) in the sample (16.7%) was slightly lower than that in the local population (18.4%). Data were not collected on the sample’s racial or educational composition, but census data reveal that the study locale is 91% White and 6% African American and one-fifth (21%) have at least a bachelor’s degree. Recent telephone surveys conducted through the lab reveal that the respondents tend to be slightly more educated than expected (with approximately one-third holding a bachelor’s degree).

Procedures

Students in an undergraduate course in social science research methods randomly called 4774 phone numbers of registered voters in a medium-sized Midwestern community in west-central Indiana. Most numbers were “bad” (e.g., disconnected, fax/modem line, etc.; N = 2117) or resulted in “no contact” with a qualified adult resident (e.g., no answer, busy signal, no adult present, etc.; N = 613). When someone answered, the caller would indicate that he/she was telephoning from the survey research lab in the psychology department at a local university. Callers would confirm that the potential respondent was 18 and seek verbal consent to continue with the survey. In total, 2044 calls that were “answered” (including immediate hang ups) yielded 854 completed interviews, a 41.8% compliance rate. Three hundred fifteen individuals (36.9%) were not in a romantic relationship; they were asked questions about a best friendship, if any; these data are not included in the current report.

Measures

The relationship survey consisted of several parts.

Self-expansion

As seen in Table 1, respondents in longer lasting relationships generally reported less self-expansion within their relationship. This association was verified in the first regression analysis, β = −.18, p < .001; however, this effect was reduced by half, and not significant, when the four covariates, age, sex, cohabitation status, and co-parenting status, were included, β = −.09, ns. But despite a significant model, R ² = .05, p < .001, none of the covariates showed significant effects either: β = −.10, ns, for age; β = .08, ns, for cohabitating with partner; and β = −.08, ns, for having children with partner—although there was a marginally significant sex difference, β = .08, p < .10, with men reporting higher rates of self-expansion. This outcome likely reflects multicollinearity among the predictors as age and relationship length were strongly correlated (r = .76, p < .001). The quadratic term for relationship length, β = .02, ns, did not significantly improve the model.

Eros

As seen in Table 1, romantic love varied little across the life span of participants’ relationships. Romantic love was unrelated to relationship length both in the initial model, β = .01, ns, and after the addition of the covariates, β = .06, ns. Nor were any of the covariates related to romantic love: β = −.06, ns, for age; β = .02, ns, for sex; β = −.06, ns, for cohabitation status; and β = .05, ns, for co-parenting status. However, self-expansion was strongly predictive of romantic love, β = .37, p < .001, even after controlling for these covariates. As seen in Table 2, self-expansion was significantly associated with romantic love at every stage of relationship, and the addition of an interaction term to the regression analysis offered no evidence that this association was moderated by relationship length, β = −.03, ns. There was, however, a significant quadratic effect for relationship length, β = .12, p < .05. This implied that the trajectory of romantic love was curvilinear across time in relationship. Further probing revealed that—as seen in Table 1—romantic love dropped gradually across relationships for the first 20+ years (vertex = 22) but then rebounded in longer lasting relationships.

Mania

As seen in Table 1 and confirmed via regression, obsessive love dropped in long-lasting relationships, as expected, β = −.27, p < .001, although this association was not significant when covariates were added to the model, β = −.12, ns. Obsessive love was significantly predicted by respondent age, β = −.22, p < .01, and sex, β = .14, p < .01 (with men reporting greater levels of obsessive love) but not cohabitation status, β = −.00, ns, nor co-parenting status, β = .03, ns. Self-expansion in a relationship was also significantly related to obsessive love, β = .15, p < .01. Although there was no evidence of a quadratic component of relationship length in this trajectory, β = −.02, ns, the interaction of self-expansion and relationship length approached significance, β = −.08, p = .06. Further probing revealed—as evident in Table 2—self-expansion was positively associated with obsessive love early in the relationship, but this effect generally diminished over time (except for those married 20–30 years—which may reflect the particular cut points used in creating the table).

Storge

As seen in Table 1, companionate love appeared to increase in longer relationships, β = .15, p < .01, but this effect was not significant when the covariates were included in the model, β = −.05, ns. Respondent age was positively related to levels of companionate love, β = .28, p < .001, but respondent sex, β = .07, ns; cohabitation status, β = −.08, ns; and co-parenting status, β = .04, ns, were not. Nor was self-expansion related to companionate love, β = −.04, ns, at any relationship duration (see Table 2) as confirmed by a nonsignificant interaction term, β = .01, ns. The quadratic term for relationship length, β = −.00, ns, also failed to achieve significance.

Implications for the self-expansion model

According to the self-expansion model (Aron & Aron, 1986), as intimacy grows and a partner is fully included in the self, self-expansion opportunities within the relationship diminish. This is consistent with the pattern obtained in the current study. Respondents in the longest lasting relationships reported lower levels of self-expansion than those in relatively newer relationships, although collinearity makes it unclear whether this is independent of age/aging. Contrary to expectations, however, there was no parallel drop in romantic love as a function of time in relationship or participant age. Although romantic love tended to drop with increasing time in relationship early on, it appears to rebound later, and people who had been with their partner for over 40 years reported just as much romantic love toward their partner as those just a few years into their relationship. Although these diverging trends seem inconsistent with the self-expansion model, self-expansion experiences were positively associated with romantic love no matter how old the relationship, just as expected by the self-expansion model. How is it that romantic love is maintained at high levels despite diminishing levels of self-expansion, given a positive association between them? Several explanations seem possible.

First, romantic love may have causes besides self-expansion with one’s partner. For instance, other relationship factors may counter the decline that would be expected with reduced self-expansion opportunities in long-term relationships (Fingerman & Charles, 2010). Perhaps early retirement or children leaving home permits partners to experience pleasure in each others’ company while carrying out the mundane activities of life that do not engender opportunities for self-expansion (Gorchoff, John, & Helson, 2008). Alternatively, even as self-expansion opportunities through one’s partner diminish in long-lasting relationships, other opportunities for self-expansion emerge that may positively impact attraction for one’s partner and relationship. For instance, one’s grown children or grandchildren may become routes for continued self-expansion that are both independent of one’s relationship with a partner (as measured here) and intimately connected to it. One might speculate that an avocation or other outside source of self-expansion might also generate positive affect—perhaps even passion—that may be transferred to a supportive partner who does not themselves participate.

A second possibility is that the affective benefit of self-expansion may change over age or time in relationship. There is no a priori reason why self-expansion in a mature relationship should generate the same level of positive affect as in a youthful relationship. In fact, it seems possible that small changes to a fully developed self (that includes one’s romantic partner) could evoke stronger responses than larger changes in a still developing self. Consistent with this, some researchers have observed that older adults enhance the significance of new love experiences through constructing a sense of their uniqueness from prior experiences in life (Ben-Ari, Lavee, & Gal, 2006); whether this pattern of “enhancement” occurs with self-expanding events in long-term relationships is mere speculation. However, if it does, lower levels of self-expansion would be needed to maintain higher levels of romantic passion in later life, which is consistent with the obtained results.

Finally, it is also possible that there is no decline in self-expansion in long-lasting relationships. Because of the cross-sectional nature of the current data, it is impossible to be certain that the lower self-expansion reported by participants from longer lasting relationships really reflects a change in their experience relative to when their relationships were newer; instead, it could be that this pattern of reports reflects the experience of participants from different cohorts. Some authors (e.g., Twenge & Campbell, 2008) have argued that American society is increasingly emphasizing the “self,” and this may have resulted in greater emphasis on close relationships as a source of self-expansion. As a result, the absolute levels of self-expansion reported by those from relatively young relationships may reflect their different experience (self-expanding through their relationships) relative to those in older cohorts (who may find self-expansion in outside activities) even as the association between self-expansion and romantic love remains the same.

Like self-expansion, the obsessive component of passionate love seemed to decline in longer lasting relationships, although this seems linked to aging rather than time in relationship, per se. While self-expansion experiences were positively associated with this component of passionate love as implied by the self-expansion model, unexpectedly, this association was most evident in relatively young relationships and seemed to diminish over time. Perhaps obsession is necessary to sustain interactions with a partner long enough for self-expansion to be initiated, but once inclusion of the other is underway, one’s focus can be released to explore other sources of expansion. Alternatively, this association may represent an attempt to jealously protect the potential for self-expansion in a partner that has not yet been realized, which may dissipate once it has (Lewandowski, 2003 similarly notes the affective importance of the potential for self-expansion). Understanding the parameters of this correlation is important because unlike romantic love, obsessive love is negatively associated with relationship satisfaction (Graham, 2011).

Aging in relationships

Although the motivation of this study was to examine the temporal course of self-expansion in relationships and its association with measures of passionate love, other trends in the data also merit consideration.

There was no evidence of a precipitous drop in passionate love in the first years of a relationship, but this may reflect the relatively small number of respondents in relationships of less than 2 years. On the other hand, a large number of respondents were in relationships of 20 years or longer, and many, much longer, which may have enabled detection of the “rebound” in romantic love at older ages. While there was a downward trend in romantic love in relationships from 0 to 20 years, the trend was positive in longer lasting relationships. This finding is relatively unique, and its causes are not immediately obvious. As noted above, family and life cycle transitions (e.g., retirement, children leaving, etc.) or other factors may allow older partners to enjoy each other more (Fingerman & Charles, 2010; Gorchoff et al., 2008) and alter the experience and trajectory of romantic love. It is also interesting to speculate whether this trend might be related to other changes in later life. For instance, researchers have noted a general increase in happiness and well-being at older ages, which could be the underlying cause (Keyes, Shmotkin, & Ryff, 2002; Weiss, King, Inoue-Murayama, Matsuzawa, & Oswald, 2012); alternatively, it is worth exploring whether changes in romantic love contribute uniquely to general well-being in older adults. This finding highlights the importance of adequately representing older adults and of testing for nonlinear patterns in relationship studies.

There was also a tendency for older respondents to report more companionate love toward their partners than younger respondents. Although one must be cautious in interpreting responses to this single-item measure, especially in light of Graham’s (2011) meta-analysis which found that companionate love (“practical friendship”) was unrelated to relationship length or satisfaction, other researchers (e.g., Bierhoff & Schmohr, 2004) have reported increases in “storge” in older relationships. It may surprise some that there was no significant correlation between companionate love and self-expansion experiences in this study as it may seem logical that romantic love (which is correlated with self-expansion experiences) eventually becomes companionate love (Aron et al., 2008; Sprecher & Regan, 1998). However, if such a transition occurs, the observed relationships are likely to be lagged by an unknown temporal parameter and difficult to see without an additional measure of “achieved” expansion or integration (perhaps the Inclusion of Other in Self scale Aron et al., 1992) to bridge these experiences. Clearly, more research is needed to understand this pattern.

Limitations

Although this study included a larger and potentially more representative sample of American adults than is typically obtained in relationship studies, it must be noted that the Midwestern population it represents is itself a culturally homogenous group, showing relatively little of the variability that might be observed in a larger, urban area. Given cultural and subcultural variability in the nature and importance of the self, it would be especially interesting to explore these relationships in a more diverse cultural group or in a locale that is not as self-focused as the United States and where passion is not seen as a basis for marriage.

While the use of a telephone survey methodology enabled collection of data from a broad range of community members, telephone assessment also has limitations, perhaps the most salient of which is the reliance on brief measures. Although the items used were carefully selected to represent established constructs, there is almost certainly a loss of precision and increase in error associated with such brief measures. For instance, the wording of the single item to represent companionate love, “Your love for your partner is really ‘just’ a deep friendship” may have inappropriately limited endorsement by respondents who also felt passion toward their partners. It is reassuring that the factor structure of the measures were consistent with those reported in other samples using larger versions of the measures. Caution is still warranted in drawing comparisons between companionate and passionate love types and in interpretation of nonsignificant findings despite a moderate large sample size.

It is also important to recognize the limitations posed by the cross-sectional nature of the data. Although the observed patterns may represent changes that typically occur in relationships across time, there are a number of potential confounding factors. For instance, the very large correlation between participants’ age and length of relationship obscured the role of time in relationship versus age as an explanation for the trajectory of self-expansion. In addition, as mentioned above, observed trends may not represent “changes” in participants’ life or relationship experiences but differences in the experiences of people who began their relationships at different times over the last 60+ years. A longitudinal design is necessary to begin to disentangle these possibilities.

One further caution is in order. Although the current results support others (e.g., Acevedo & Aron, 2009) in questioning the common wisdom that passion and romance necessarily fade, it is important to observe—as have others (e.g., Hatfield et al., 2008)—that those whose passion faded the most are likely missing from the current study (as their relationships have probably ended). This could easily explain the absence of an expected downward trend in passionate love across longer lasting relationships, but there is no obvious reason to suspect that would have affected the significant nonlinear trends and associations between self-expansion and passionate love reported here. As a result, while one cannot conclude from the current findings that romantic love does not weaken over time, one can conclude that it does not have to.