There is more than one pathway to romance, but relationship science does not reflect this reality. Our research reveals that relationship initiation studies published in popular journals (Study 1) and cited in popular textbooks (Study 2) overwhelmingly focus on romance that sparks between strangers and largely overlook romance that develops between friends. This limited focus might be justified if friends-first initiation was rare or undesirable, but our research reveals the opposite. In a meta-analysis of seven samples of university students and crowdsourced adults (Study 3; N = 1,897), two thirds reported friends-first initiation, and friends-first initiation was the preferred method of initiation among university students (Study 4). These studies affirm that friends-first initiation is a prevalent and preferred method of romantic relationship initiation that has been overlooked by relationship science. We discuss possible reasons for this oversight and consider the implications for dominant theories of relationship initiation.
Relationship scientists’ empirical attention has not been evenly distributed along the trajectory of romance (Eastwick et al., 2019). For example, researchers have devoted considerable attention to studying the initial spark of attraction that kindles between two strangers meeting for the first time. Or, to be perfectly accurate, researchers have devoted considerable attention to studying the spark of attraction that kindles when someone views a photograph, reads a brief biography, or views a list of traits that may be possessed by a potential romantic partner. However, as Eastwick and colleagues point out, a lot can happen between a spark of attraction and maintaining a committed romantic relationship. Unfortunately, because “[t]he initial attraction literature does not intersect empirically with the literature on established romantic relationships” (Eastwick et al., 2019, p. 2), researchers remain mostly in the dark about what those processes might be.
Pathways to Romance
Relationship scientists have long understood that there are at least two kinds of intimacy (e.g., Berscheid, 2010; Guerrero & Mongeau, 2008). One is friendship-based intimacy, which is a cognitive and emotional experience comprising psychological interdependence, warmth, and understanding, related to the companionate love that nurtures long-term intimate bonds. The other is passion-based intimacy, which is a primarily emotional experience comprising romance and positive arousal, related to the passionate love that typifies novel, and often sexual, relationships. The dominant, but heterosexist, dating script proposes that men’s passionate desire sparks the initial interaction between potential romantic partners and then passion-based intimacy and friendship-based intimacy continue to develop over time in tandem.
However, in her biobehavioral model of sexual orientation, Diamond (2003) convincingly argues that while emotional affection (i.e., friendship-based intimacy) and sexual desire are distinct, the biobehavioral links between systems are bidirectional. Thus, even though sexual desire can precede and even nurture friendship-based intimacy, as the dating script prescribes, the opposite can also occur: Two people can become friends, develop a deep friendship-based intimacy and then begin to experience sexual desire at some future point in time. Now, the dating script might suggest that such friendships are not truly platonic, and concealed passionate desire is the true motivation behind such bonds. After all, some 30%–60% of (presumably heterosexual) cross-sex friends report at least moderate sexual attraction for one another (e.g., Halatsis & Christakis, 2009; Kaplan & Keys, 1997). Yet the empirical evidence is clear that friendship-based intimacy can precede and even nurture passion-based intimacy (see Rubin & Campbell, 2012). When this happens, the friends may decide not to act on their passion (Bleske-Rechek et al., 2012), or they may form a “friends-with-benefits” relationship, where they engage in sexual activity with rules to limit emotional attachment (Mongeau & Knight, 2015). Yet while friends-with-benefits relationships are very common among young people, only a very small proportion ever transition to a traditional romantic relationship (Bisson & Levine, 2009; Machia et al., 2020).
Thus, most friendships that eventually transition to romance must follow a different path. Indeed, as Diamond (2003) reviews, the friends-first pathway to romance is well-documented among people who experience same-gender/sex1 attractions, even among people who self-identify as heterosexual. Furthermore, Eastwick and colleagues (2019) note that the few studies examining the trajectory of early romance suggest that people often know one another for months or even years before they officially enter couplehood. Although it was not the primary focus of the research, two longitudinal studies of romantic relationships between men and women report that a meaningful proportion began as friendships (Eastwick et al., 2019; Hunt et al., 2015). Together with Diamond’s (2003) research, these longitudinal studies suggest that romantic and sexual attraction can blossom within long-standing platonic friendships between people of all genders, and sometimes those feelings can lead to romantic couplehood.
![](https://static.wixstatic.com/media/9afd33_65555bbbe46547a181d0cd4f305af5bd~mv2.png/v1/fill/w_960,h_1280,al_c,q_90,enc_auto/9afd33_65555bbbe46547a181d0cd4f305af5bd~mv2.png)
The Current Research
We present four studies designed to quantify the extent of researchers’ potential neglect of friends-first initiation as well as the prevalence of and people’s preferences for this form of initiation. In Studies 1 and 2, we systematically code the literature on relationship initiation to determine how often researchers study dating versus friends-first initiation. In Study 3, we seek to establish the prevalence of friends-first initiation with a meta-analysis of seven studies that we have conducted, involving nearly 1,900 participants. We also explore group differences in the prevalence of friends-first initiation (i.e., gender, age, sample population [students vs. MTurk workers], education level, ethnicity, gender composition of the couple), and we explore the prevalence of friends-with-benefits relationships among now-married friends-first initiators. In Study 4, we delve deeper by exploring how long university students were friends prior to couplehood and whether they entered those friendships to facilitate an eventual romance. Participants also report the “best way to meet a dating or romantic partner,” allowing us to assess their preference for friends-first initiation.
If our descriptive and exploratory results reveal that friends-first initiation is a prevalent and preferred method of relationship initiation that is relatively overlooked in relationship science, it will suggest that researchers need to revisit the validity of dominant models of relationship formation, all of which were devised based on research that likely focuses almost exclusively on dating initiation, and all of which may operate very differently during friends-first initiation. For example, one of the only in-depth studies of friendship initiation to date revealed that assortative mating for physical attractiveness was much weaker among friends-first initiators compared to dating initiators (Hunt et al., 2015). Furthermore, although first dates involving (presumably heterosexual) women and men typically follow gender-role prescriptions, expectations for first dates are more egalitarian during friends-first initiation, which may alter the power structure of developing relationships (Cameron & Curry, 2020; see also Rose, 2000). Friendship-based intimacy is also the foundation of long-lasting romantic bonds (VanderDrift et al., 2016), and thus understanding how and when people transition from friendship to romance may help researchers to understand the social–psychological foundations of strong and satisfying romantic relationships. In addition, exploring the transition from friendship to romance reveals the messy reality of relationship initiation, which belies the orderly sexual scripts that dominate Western culture. We will return to these issues in the Discussion, but in general, we suspect that by overlooking friends-first initiation, psychologists may have a surprisingly limited understanding of how people actually form romantic relationships.
![](https://static.wixstatic.com/media/9afd33_9f53e586155f4a6aa357e1bcb171ab2e~mv2.png/v1/fill/w_980,h_703,al_c,q_90,usm_0.66_1.00_0.01,enc_auto/9afd33_9f53e586155f4a6aa357e1bcb171ab2e~mv2.png)
Seventy-nine percentage (n = 85) of the sampled publications concerned romantic relationship initiation (Table 1). Of these, 74% (n = 63) concerned dating initiation while only 8% (n = 7) focused on friends-first initiation (Table 2). A further 9% (n = 8) concerned both types of initiation, and if we include these articles in both counts, then 84% of articles concerned dating initiation while only 18% concerned friends-first initiation. These results suggest that researchers largely overlook friends-first initiation and overwhelmingly focus on dating initiation in their empirical study of relationship initiation (articles are listed in the OSM).
![](https://static.wixstatic.com/media/9afd33_244b6404ace0484aa88cd0be361d758d~mv2.png/v1/fill/w_980,h_719,al_c,q_90,usm_0.66_1.00_0.01,enc_auto/9afd33_244b6404ace0484aa88cd0be361d758d~mv2.png)
![](https://static.wixstatic.com/media/9afd33_cae8b19e745744e2a0452014e6592a5a~mv2.png/v1/fill/w_980,h_551,al_c,q_90,usm_0.66_1.00_0.01,enc_auto/9afd33_cae8b19e745744e2a0452014e6592a5a~mv2.png)
A meta-analysis of proportions using an exact binomial-normal model and logit transformed proportions (e.g., Hamza et al., 2008) with the metafor package in RStudio (metafor version 2.4-0; Viechtbauer, 2010; RStudio version 3.6.2) revealed a median proportion of 0.66 (95% confidence interval [CI] = [0.59, 0.72]), and an estimated average log-odds of 0.66 (SE = 0.14), 95% CI [0.38, 0.95], z = 4.62, p < .001, indicating that the majority of romantic relationships began as friendships (see Figure 1 for forest plot and the OSM for R code). There was also significant heterogeneity among studies, Wald (6) = 31.01, p < .001, I2 = 86%, H2 = 7.14, τ2 = 0.12, indicating that effect sizes varied across studies perhaps due to the use of different measures. These results reveal that friends-first initiation is a common form of relationship initiation.
General Discussion
Our results reveal that psychologists have largely overlooked the most prevalent and desirable form of relationship initiation. Even though two thirds of the nearly 1,900 participants in the studies that we meta-analyzed in Study 3 reported friends-first initiation, and even though 47% of the university age participants in Study 4 claimed that friends-first initiation is the best way to initiate a relationship, just 18% of the studies that we located in our literature search actually focused on this method of initiation. Notably, our impression is that many of these studies covered friends-first initiation in a brief or peripheral manner. Given the paucity of research on friends-first initiation, it is not surprising that the textbooks we coded only cited two articles that focused on friends-first research at all, and these works exclusively focused on friends-with-benefits relationships. This means that the field of close relationships has only a partial understanding of how romantic relationships actually begin.
There are certainly flaws in our research that should be addressed in future studies. Our research concerning the prevalence of friends-first initiation was based on retrospective reports. Such reports are easy to collect, but they can be biased by subsequent experience, and this threat to validity may be particularly salient for emotionally charged experiences like romance (Holmberg et al., 2004). Longitudinal, prospective studies may be better suited to studying friends-first initiation. In addition, although the samples we included in Study 3 lived in different regions of Canada and the United States and comprised both younger and older adults, our samples were still relatively WEIRD (i.e., Western, Educated, Industrialized, Rich, & Democratic; Henrich et al., 2010) and most samples did not include singles. Future research should examine cultural differences in the prevalence of friends-first initiation and other forms of initiation that may not be commonly recognized in the West, but which may be similarly overlooked by extant scientific theories and data. Further, although our analyses in Study 3 suggested that friends-first initiation is more common among same gender/queer couples than among couples that include a man and a woman—perhaps due to group differences in the size of the available dating pool, differing scripts concerning intimacy and communication, and fluid understandings of gender, among other reasons (e.g., Rose, 2000)—our sample size for the former group was very small (N = 84 across four samples; see the OSM), and we did not explicitly assess sexual orientation. Although our results are not definitive, they do support prior observations made about same-sex romantic relationship formation (Diamond, 2003). Nevertheless, future research examining the prevalence of various relationship-initiation strategies should include all sexual orientations. Research should also examine whether friends-first initiation is preferred among older adults, as our sample in Study 4 comprised university students. Finally, we did not define “friendship” for any of our participants, so our results may be biased by participants’ ability to self-define a relationship that lacks a precise and shared cultural definition to begin with (e.g., VanderDrift et al., 2016). Future research should seek to document the characteristics of friendships that do and do not lead to romance and to ensure that our prevalence rate is not potentially inflated by some participants’ excessively broad interpretation of friendship.
But to achieve these important goals and develop a science of relationship initiation that truly reflects people’s behavior, researchers may need to take a cold, hard look at the reasons why the field has overlooked friends-first initiation in the first place (and yes, we include ourselves in this critique). As we explained in the introduction, it is difficult to study social–psychological phenomena that occur spontaneously and in private, and it is easier to use experimental paradigms that enhance scientific control. Yet researchers’ preference for these methods may have shaped the very questions we think to ask, a kind of “tail wags the dog” situation that may have diverted attention away from friends-first initiation. Thus, researchers and funding agencies need to invest in more longitudinal studies that offer the possibility of capturing different types of relationship initiation as they spontaneously occur.
Moreover, as we explained in the introduction, implicit heterosexist biases hinder relationship science (Rose, 2000) and that may help to explain researchers’ relative neglect of friends-first initiation. For example, despite convincing evidence that passion-based intimacy can arise from friendship-based intimacy among same-gender friends (e.g., Diamond, 2003), it may not have occurred to researchers that such a thing could also happen in platonic friendships between heterosexual men and women. Moreover, if people assume that men and women cannot be platonic friends because sexual attraction inevitably gets in the way, and if researchers assume that everyone desires and prioritizes romantic relationships over friendships and singlehood (but see Bay-Cheng & Goodkind, 2016; Fisher & Sakaluk, 2020; Fisher et al., 2021), it may be difficult to conceive of the possibility that heterosexual men and women might maintain a platonic friendship for months or even years, like our Study 4 participants, before romantic feelings start to blossom. Interrogating and overcoming these and other heterosexist assumptions about relationships may be the first step to developing a science of relationship initiation that truly reflects the full diversity of human experience.
The gulf between the fields’ excessive scientific focus on dating initiation and people’s frequent lived experiences of friends-first initiation also has important implications for theories of relationship formation and maintenance. Researchers may need to revisit the validity of dominant models of relationship formation, including risk-regulation theory (Cameron et al., 2010; Stinson et al., 2015), sexual strategies theory (e.g., Eastwick et al., 2018), and assortative mating (e.g., Fletcher et al., 2000; Hoplock et al., 2019), all of which were devised by studying dating initiation, and all of which may apply differently, or not at all, to the process of friends-first initiation (see Hunt et al., 2015). Moreover, researchers should examine whether people exhibit systematic preferences for one type of initiation or another, and whether psychological variables like attachment (Mikulincer & Shaver, 2020), sociosexuality (e.g., Gangestad & Simpson, 1990), life history (e.g., Belsky, 2012), or personality (e.g., McNulty, 2013) predict that preference. They may also need to examine whether these same variables moderate the success of each type of initiation, and whether such variables moderate the trajectory of relationships that form via dating or friends-first initiation. As such, studying friends-first initiation may be a fruitful enterprise that not only promises to expand extant theories of relationship initiation, but which also promises to shed light on new aspects of relationship initiation that could shift our understandings of how romantic relationships begin and progress.
Stinson, D. A., Cameron, J. J., & Hoplock, L. B. (2022). The Friends-to-Lovers Pathway to Romance: Prevalent, Preferred, and Overlooked by Science. Social Psychological and Personality Science, 13(2), 562-571. https://doi.org/10.1177/19485506211026992
Comments