This is an open access article distributed under the terms of the Creative Commons
Attribution License (Social scientists have long wondered why and how people get to like one another. This fascination led to a large body of work—spanning decades—discovering various predictors of attraction. Such research has largely focused on examining the factors that preceded or led to the emergence of attraction in established relationships. Perhaps one of the most common methodologies for addressing these questions has been the “mate selection paradigm,” in which people rate the importance of a variety of characteristics that they desire in a potential close partner. This type of work has shown some consistent patterns in the traits that people desire most. For example, Lippa’s (2007) analysis of over 100,000 people suggested that some of the traits that both men and women desire most include a sense of intelligence, honesty, and kindness.
Some of the earliest work on the question of why people get to like one another revealed that married couples tended to live in proximity to one another prior to the issue of their marriage license (Bossard, 1932; Davie & Reeves, 1939). Similarly, social psychologists demonstrated that the principles of propinquity and its effects on attraction applied to friendship formation (Festinger, Schachter, & Back, 1950). As this line of work expanded, researchers identified further factors that affect the emergence of attraction. In their seminal “computer dance” study, Walster, Aronson, Abrahams, and Rottman (1966) demonstrated that the strongest predictor of attraction in students who were blindly paired with another student to attend a dance was the target’s physical attractiveness. Another strong predictor of attraction that researchers had identified was similarity. One such demonstration was Byrne, Ervin, and Lamberth’s (1970) “Coke date” study where participants went on a blind date with an opposite-sex other (the “Coke” reflects the sweetened carbonated beverage that participants were able to purchase with money given to them by the researchers). Byrne and colleagues not only replicated Walster et al.’s (1966) findings that physical attractiveness predicted attraction, but demonstrated that similarity uniquely predicts attraction as well. This collective early research cemented the consistent positive effect of similarity on attraction (Byrne, 1997).
People may be aware that similarity is important in establishing novel relationships. For example, when evaluating partner preferences for several types of relationships with varying nature of commitment (e.g., casual sex relationships, marriage, same-sex friends), people considered traits such as similarity in backgrounds, attitudes, and leisure skills as being some of the most important precedents for liking someone (Sprecher & Regan, 2002). Not all similarity, however, may be created equal. Researchers have identified two general types of similarity: actual and perceived. Whereas actual similarity reflects the correlation between persons’ attitudes, personalities, or other variables, perceived similarity reflects the degree to which one believes that they are similar to another person. In their meta-analysis, Montoya, Horton, and Kirchner (2008) sought to examine whether these two types of similarity differ in the degree to which they beget attraction. Their results revealed an interesting finding: perceived, rather than actual similarity was a better predictor of attraction.
More recent experimental investigations have demonstrated the same result. Sprecher, Treger, Fisher, Hilaire, and Grzybowski (2015) found in a social interaction study that actual similarity in dyads’ attitudes was generally unrelated to attraction following self-disclosure, whereas perceived similarity and attraction were positively associated. Likewise, research using a speed dating methodology (i.e., an event in which people participate in a series of brief “mini-dates” with a number of others) found that perceived similarity superseded actual similarity in predicting post-date liking (Tidwell, Eastwick, & Finkel, 2013). Interestingly, the positive effects of similarity on liking is not limited to interpersonal relationships. Kwon, Ha, and Im (2016), for example, found that people enjoyed their shopping experience at a mall more if they found themselves to be similar to other shoppers.
Why Does Similarity Beget Attraction?
Although researchers have long established that similarity predicts attraction, “…the explanation for this effect has been the subject of much debate” (Montoya & Horton, 2013, p. 65). Some theoretical explanations behind this link, however, have emerged throughout the line of this work. Byrne, Clore, Griffitt, Lamberth, and Mitchell (1973) suggested that similarity reinforces one’s worldview by showing that another agrees with them, in turn leading to positive affective reactions, citing work such as Byrne and Nelson (1965) and Clore and Baldridge (1968) who found that the degree of agreement between a participant and a bogus stranger predicted liking for the stranger.
Alternatively, attraction stemming from similarity may be an artifact of knowledge one has about another. That is, the more information one may have about another, the more favorability may be found in that information (i.e., information salience; Kaplan & Anderson, 1973). Because people tend to view themselves favorably, seeing similarity in another leads to similar levels of favorability (i.e., attraction). Indeed, more recent research has suggested that generally, there is a positive correlation between familiarity (i.e., the amount of information one knows about another) and attraction (Reis, Maniaci, Caprariello, Eastwick, & Finkel, 2011; Sprecher, Treger, & Wondra, 2013). This idea is perhaps further demonstrated by the phenomenon of implicit egotism, which reflects unconscious positive evaluation of stimuli that even trivially refer to oneself, such as sharing a similar name or birthday (Jones, Pelham, Carvallo, & Mirenberg, 2004). Similarly, sharing even small subjective experiences (e.g., enjoying a meal to a similar extent), what some researchers label as I-sharing (Pinel, Long, Landau, Alexander, & Pyszczynski, 2006), facilitates attraction between two strangers. In their comparison between the reinforcement and the information salience models of the similarity-attraction link, Montoya and Horton (2013) concluded via a meta-analysis that the latter model may be stronger than the former.
Further theoretical work suggests that similarity-induced attraction is grounded in reciprocal liking, or one’s perception that another person likes them (Condon & Crano, 1988)—this reciprocal liking effect in and of itself stems from one’s perception that the other person possesses positive or warm intentions towards them (Montoya & Insko, 2008). In a study of romances and friends, for example, Baxter and West (2003) found that perceptions of similarity open the door for perceptions of positive outcomes such as warmth, fun, and support, all of which lead to attraction. Further perspectives would suggest that interacting with similar others is intrinsically rewarding and enjoyable (Berscheid & Walster, 1969), and enjoyment of an interaction with another in and of itself predicts attraction (Treger, Sprecher, & Erber, 2013).
The Current Study
The link between general (perceived) similarity and attraction has been well-established for decades of scientific work (Montoya & Horton, 2013; Montoya et al., 2008). Still, similarity as a concept is broad and may entail many dimensions. Although people may be similar “on average,” they may differ in certain aspects such as preferences in music or preferences in technology use that may be more important than others. Some of such domains may be considered by individuals necessities or “deal-breakers.” Even if general similarity is present, perhaps (for example) if one’s tastes in music do not align with a potential other’s, they may not desire to form a relationship (e.g., Li, Bailey, Kenrick, & Linsenmeier, 2002). In this research, we contribute to the current knowledge of the role of similarity on attraction by considering whether similarity in some domains may be more important in predicting attraction than may similarity in other domains. In other words, we wondered whether people may weight certain similarity domains more so than others when establishing attraction.
We considered 18 domains of similarity. These domains are based on certain attitudes (e.g., political views) and popular pastimes that people may enjoy (e.g., music, video games). We were particularly interested in whether certain domains are more important than others when considering a new partner for three types of relationships: friendship, casual/short-term dating, and long-term dating. Furthermore, we also considered whether men and women differ in any preference for similarity domains. In this research, we tested one hypothesis and four research questions. Because people are generally more selective of long-term versus short-term partners and friendships (Sprecher & Regan, 2002), we proposed the following hypothesis:
Hypothesis 1 (H1): Participants will, overall, be more selective (place more emphasis overall) in long-term relations vs. (a) short-term relationships and (b) friendships.
Certain effects in which we were interested have not been addressed prior, and do not necessarily entail a theoretical foundation for testing particular hypothesis. We thus further considered four research questions:
Research Question 1 (RQ1): Which similarity domains will participants report as being most important?
Research Question 2 (RQ2): Will the importance of similarity domains differ across the three relationship types?
Research Question 3 (RQ3): Will men and women differ in the domains of similarity they find important?
Research Question 4 (RQ4): Will men and women differ in the importance of the similarity domains across the three types of relationships?
Method
Participants and Procedure
A total of 250 Amazon.com’s Mechanical Turk (MTurk) workers self-selected to participate in this study by clicking on a link through the MTurk platform. We constrained the sample to be United States residents only given that a large body of work on mate preferences used United States samples. We could not analyze data from 29 participants in our tests of hypotheses. Two participants did not disclose their sex, which prohibited us from using them in our analyses, given that sex is a between-subjects variable; and we could not analyze 27 participants because they did not provide enough responses to perform our within-subjects tests (i.e., not providing responses to one or more similarity domains). Thus, the final sample size was 221 (97 men). Approximately 28% of the participants reported to be romantically-unattached, 5% reported to be in a short-term relationship, 62% reported to be in a long-term relationship, with the remainder of the sample reporting “other.” The average age of the sample was 36.93 (SD = 12.58, range = 18, 80). We paid participants $0.10 to partake in this study.
Materials
We presented participants with 18 domains of similarity, and asked them to rate the degree to which each domain is important to them when forming new friendships, short-term romantic relationships, and long-term romantic relationships using the scale 1 = Not important at all; 3 = Somewhat important; 5 = Quite important; and 7 = Very important.
Results
Descriptive Analyses
See Table 1 for overall means of participants’ reports of importance for each similarity domain, ordered from most to least important. Participants believed that similarity in political views is the most important domain on this list when considering a new relationship partner (this was the only mean that was not different from the midpoint of the scale, t = -0.46, p = .6443). The remaining four of the five most important domains were, in order: career goals, food preferences, travel desires, and music preferences.
Table 1
Importance of Each Domain of Similarity.
| Similarity Domain | Overall Mean | SD | Men | SE | Women | SE | g | p |
|---|---|---|---|---|---|---|---|---|
| Political Views | 3.95 | 1.76 | 3.79 | 0.18 | 4.20 | 0.16 | -0.11 | .087 |
| Career Goals | 3.71 | 1.43 | 3.60 | 0.15 | 3.70 | 0.13 | -0.08 | .619 |
| Food Preferences | 3.45 | 1.45 | 3.37 | 0.15 | 3.56 | 0.13 | -0.07 | .348 |
| Travel Desires | 3.45 | 1.47 | 3.32 | 0.15 | 3.50 | 0.13 | -0.15 | .372 |
| Music Preferences | 3.41 | 1.56 | 3.32 | 0.16 | 3.53 | 0.14 | -0.10 | .308 |
| Movie Preferences | 3.26 | 1.44 | 3.21 | 0.15 | 3.33 | 0.13 | -0.02 | .546 |
| Cultural Background | 3.11 | 1.59 | 2.95 | 0.16 | 3.15 | 0.14 | -0.11 | .346 |
| TV Show Preferences | 2.97 | 1.46 | 2.87 | 0.15 | 3.09 | 0.13 | -0.06 | .279 |
| Exercise Preferences | 2.87 | 1.44 | 2.99 | 0.15 | 2.73 | 0.13 | 0.21 | .184 |
| Book Preferences | 2.80 | 1.43 | 2.66 | 0.15 | 2.84 | 0.13 | -0.10 | .360 |
| Technology Use | 2.66 | 1.32 | 2.76 | 0.13 | 2.55 | 0.12 | 0.19 | .235 |
| Social Media Use | 2.47 | 1.26 | 2.47 | 0.13 | 1.47 | 0.11 | 0.001 | .989 |
| Beverage Preferences | 2.45 | 1.43 | 2.55 | 0.14 | 2.40 | 0.13 | 0.10 | .449 |
| Number of Friends | 2.40 | 1.30 | 2.32 | 0.13 | 2.42 | 0.12 | -0.06 | .574 |
| College Major | 2.39 | 1.30 | 2.44 | 0.13 | 2.32 | 0.12 | 0.07 | .489 |
| Video Game Preferences | 2.23 | 1.50 | 2.46 | 0.15 | 2.03 | 0.13 | 0.33 | .031 |
| Computer Brand Preferences | 1.62 | 1.06 | 1.73 | 0.10 | 1.49 | 0.09 | 0.27 | .093 |
| Cell Phone Brand | 1.60 | 1.08 | 1.74 | 0.11 | 1.46 | 0.10 | 0.26 | .060 |
Note. The domains are organized by reported importance, from the most to least important. Negative values of g reflect higher scores for women than for men.
As an exploratory test, we examined whether romantically-unattached persons differed from those in a long-term relationship in the domains of similarity. We did not include the other three relationship statuses as their samples were too small for analyses. The tests suggested that the two relationship groups did not differ in preferences for domains with the exception of three: music tastes (Ms = 3.71 [SD = 1.71] and 3.17 [SD = 1.44] for unattached persons versus those in a committed relationship, Welch’s t = 2.24, p = .0271, g = 0.35); book preferences (Ms = 3.32 [SD = 1.47] and 2.51 [SD = 1.32] for unattached persons versus those in a committed relationship, Welch’s t = 3.90, p = .0271, g = 0.59); and political views (Ms = 4.44 [SD = 1.64] and 3.74 [SD = 1.77] for unattached persons versus those in a committed relationship, Welch’s t = 2.87, p = .0271, g = 0.40). Because we were uninterested in the effects of relationship type on preferences for similarity domains, we excluded this variable from our analyses.
Tests of Hypothesis and Research Question
Overview
We tested our hypothesis and research questions using a 3 (Relationship Type) x 18 (Similarity Domain) x 2 (Sex) mixed-model ANOVA with relationship type and similarity domain serving as within-subjects variables, and sex serving as a between-subjects variable1. As shown in Table 2, main effects of relationship type and similarity domain emerged—these effects, however, were qualified by a relationship type x similarity domain, a relationship type x sex, and a similarity domain x sex interaction. We used Šidak corrections in all analyses of simple effects when probing interactions. We will now turn to a discussion of these findings.
Table 2
Results of the Relationship Type x Similarity Domain x Sex ANOVA
| Main Effects | F | df | η2p | p |
|---|---|---|---|---|
| Relationship Type | 101.57 | 2, 438 | .317 | < .001 |
| Similarity Domain | 85.93 | 17, 3723 | .28 | < .001 |
| Sex | 0.009 | 1, 219 | < .001 | .926 |
| Interactions | ||||
| Relationship Type x Sex | 9.19 | 2, 438 | .036 | < .001 |
| Similarity Domain x Sex | 2.58 | 17, 3723 | .012 | < .001 |
| Relationship Type x Similarity Domain | 15.34 | 34, 7446 | .065 | < .001 |
| Relationship Type x Similarity Domain x Sex | 1.23 | 34, 7446 | .006 | .170 |
Testing H1 and RQ4
Figure 1 displays the results of the relationship type x sex interaction, which addresses H1a, H1b, and RQ4. Follow-up simple effects analyses revealed support for H1a and H1b: both men and women believe that general similarity across the 18 domains is more important in long-term relationships than in short-term relationships and friendships (which did not differ from one another for both sexes). Still, the magnitude of these differences was not equal between men and women, which provides insight into RQ4. Specifically, women appeared to have placed greater emphasis on similarity across the domains (gs = 0.99 and 1.16 for short-term relationships and friendships, respectively) than did men (gs = 0.68 and 0.58 for short-term relationships and friendships, respectively). Thus, general similarity across the domains was most important in long-term relationships, but the importance was greater for women than it was for men.
Figure 1
Importance of similarity across all domains by relationship type for men and women.
Note. Error bars reflect the standard error of the mean, which is the value at the bottom of each bar.
Testing RQ2
The relationship type x similarity domain provides insights into RQ2 (see Table 3). Simple effects analyses revealed that although participants believed that similarity was most important in long-term relationships compared to short-term relationships and friendships, four domains emerged as exceptions: book preferences, video game preferences, computer brand preferences, and cell phone brand preferences. In the examination of book preferences, we found that each type of relationship differed from one another. Specifically, similarity in book preferences was most important in long-term relationships (gs = 0.37 and 0.52 for differences with friendships and short-term relationships, respectively); and similarity in book preferences was more important in friendships than in short-term relationships, although this difference was small enough to be deemed as trivial, g = 0.17. Participants believed that similarity in video game preferences was equal for friendships and long-term relationships, and somewhat less important in short-term relationships than the other two relationship types (gs = 0.21 and 0.26 for differences between friendships and long-term relationships, respectively). The importance of similarity in computer brand preferences was equal for friendships and short-term relationships, yet it was higher for long-term relationships versus short-term relationships; still, the small size of this difference may deem it as trivial, g = 0.17. Finally, there were no differences in similarity in cell phone brand preferences across the three types of relationships.
Table 3
Importance of Similarity in Each Domain Across Relationship Type
| Similarity Domain | Friendship | SE | Short-Term | SE | Long-Term | SE | F | p | η2p |
|---|---|---|---|---|---|---|---|---|---|
| Music Preferences | 3.26a | 0.12 | 3.17b | 0.13 | 3.84ab | 0.12 | 23.75 | < .001 | .18 |
| Movie Preferences | 3.09a | 0.11 | 3.11b | 0.12 | 3.61ab | 0.12 | 16.86 | < .001 | .13 |
| TV Show Preferences | 2.76a | 0.11 | 2.76b | 0.12 | 3.42ab | 0.12 | 31.67 | < .001 | .23 |
| Food Preferences | 3.11a | 0.11 | 3.22b | 0.13 | 4.06ab | 0.12 | 51.55 | < .001 | .32 |
| Beverage Preferences | 2.34a | 0.10 | 2.33b | 0.11 | 2.75ab | 0.11 | 14.23 | < .001 | .12 |
| Book Preferences | 2.67ad | 0.11 | 2.40bd | 0.11 | 3.18ab | 0.13 | 32.49 | < .001 | .23 |
| Travel Desires | 2.77a | 0.12 | 3.02b | 0.13 | 4.45ab | 0.12 | 134.64 | < .001 | .55 |
| Exercise Preferences | 2.46a | 0.11 | 2.63b | 0.12 | 3.48ab | 0.12 | 60.32 | < .001 | .36 |
| Video Game Preferences | 2.33e | 0.10 | 2.03eb | 0.11 | 2.37b | 0.12 | 8.64 | < .001 | .07 |
| College Major | 2.15a | 0.13 | 2.18b | 0.10 | 2.80ab | 0.11 | 24.77 | < .001 | .19 |
| Political Views | 3.70a | 0.10 | 3.67b | 0.14 | 4.61ab | 0.13 | 53.68 | < .001 | .33 |
| Technology Use | 2.46a | 0.10 | 2.46b | 0.11 | 3.04ab | 0.11 | 20.54 | < .001 | .16 |
| Career Goals | 3.08a | 0.12 | 3.17b | 0.14 | 4.70ab | 0.11 | 104.44 | < .001 | .49 |
| Number of Friends | 2.23a | 0.09 | 2.14b | 0.10 | 2.74 | 0.11 | 22.17 | < .001 | .17 |
| Cultural Background | 2.70a | 0.12 | 2.87b | 0.12 | 3.56ab | 0.13 | 40.00 | < .001 | .27 |
| Social Media Use | 2.29a | 0.10 | 2.24b | 0.10 | 2.89ab | 0.12 | 20.68 | < .001 | .16 |
| Computer Brand Preferences | 1.62 | 0.08 | 1.50f | 0.07 | 1.70f | 0.09 | 3.22 | .042 | .03 |
| Cell Phone Brand | 1.55 | 0.08 | 1.56 | 0.08 | 1.69 | 0.09 | 2.39 | .094 | .02 |
Note. Values sharing an identical subscript in each row represent a statistically-significant difference. Subscripts a, b, and c represent p < .001; subscript d represents p = .041; subscript e represents p = .005; and subscript f represents p = .031.
Testing RQ3
We also present in Table 1 separate means for men and women that were derived from the marginal means of the Similarity Domain x Sex interaction in the ANOVA described below (with Šidak corrections) that served as tests of RQ3. These analyses revealed that overall, men and women did not differ in the importance they placed on the similarity domains with the exceptions of video game preferences, which men believed to be somewhat more important than did women. Two other domains of similarity, computer brand preferences and cell phone brand preferences, were marginally-significant with small effect sizes, suggesting that men believed that similarity in these two domains are somewhat more important than did women. The low means for both sexes on this similarity domain, though, suggest that similarity in video game preferences are generally not important to either.
Discussion
Are some domains of similarity more important than are others? Researchers have long shown that people desire a partner that is (perceivably) similar to themselves (Byrne et al., 1970; Montoya et al., 2008; Montoya & Horton, 2013; Sprecher & Regan, 2002; Sprecher et al., 2015). Overall similarity, however, is a broad concept. People vary in their general interests and opinions, which may suggest that perceived similarity may be fairly malleable—similarity in some particular domains may be more important than in others. We considered this idea in this research. An analysis of preferences for similarity in 18 domains using a large sample of MTurk “workers” yielded a number of interesting findings. Replicating prior work (e.g., Sprecher & Regan, 2002), we saw that participants were most selective for long-term partners, desiring overall similarity in such relationships to a greater extent than in short-term romantic bonds and friendships. This finding may not necessarily be surprising, as a long-term (vs. a short-term) relationship represents greater investment of physical and emotional resources, along with greater commitment, and partner traits that may drive continued attraction. Indeed, given its consistent positive effects on attraction (Montoya et al., 2008), and people’s general desire for similarity in a long- versus a short-term mate (Sprecher & Regan, 2002), perceived similarity may knowingly be one of the driving forces that may “glue” a couple together (e.g., Baxter & West, 2003). Still, a relationship type by similarity domain interaction we found suggested that certain traits may be more important in some relationships over others; or alternatively, they may be completely irrelevant to attraction. (For example, it appears that people generally placed very little importance on their potential partner’s brand of cell phone or computer brand preferences).
The five most important domains in our study were political attitudes, career goals, food preferences, travel desires, and music preferences. These domains may converge to reflect one’s degree of openness to experience, a personality trait that entails facets such as imagination, tolerance, intellectualism, and open-mindedness (John & Srivastava, 1999). Thus, it is possible that, overall, people may prefer a partner with high levels of this trait. Men and women generally did not appear to differ in the emphasis they placed on similarity in the 18 domains with one exception. Men desired a partner who has similar interests in video games more than did women. Similar trends in other technological domains also emerged, albeit to a smaller extent that needs to be interpreted with caution. That is, men, compared to women, desired a partner who has similar preferences in computer and cell phone brands. Nevertheless, we emphasize that these effects were fairly small in our sample and, given their nature, unlikely to be of any real consequence on relationship quality on their own. This finding, though, reflects interesting prior work that has demonstrated that men prefer to work with “things” more so than with “people,” which women prefer more than do men (Woodcock, Graziano, Branch, Habashi, Ngambeki, & Evangelou, 2013). As such, these differences on what are, at first blush, trivial domains may be indicative of deeper differences indicated in prior research.
Why did men and women agree on their preferred similarity in most domains? Many of these preferences are not necessarily grounded in evolutionary or societal constraints, as domains such as movie preferences or social media did not aid in early humans’ survival and reproduction. Still, prior work has demonstrated sex differences in some of these domains. For example, women tend to be more active on the social media website Facebook to a greater extent than do men (McAndrew & Jeong, 2012), which may reflect prior work demonstrating that women are more likely than men to obtain their intimacy needs from their close social network (Fehr, 1996). Differences in use, however, may not reflect differences in preferences. Although women may use it more than may men, both may endorse similar attitudes towards the website. Still we did find one sex difference: video game use. Men reported that similarity in video game preferences was more important overall than it was for women. This may be unsurprising, however, as men tend to play videogames more so than do women (Ogletree & Drake, 2007). This finding is also complemented by men’s (vs. women’s) greater emphasis on similarity in computer and cell phone brand preferences, although these differences did not reach statistical significance, yielding small effect sizes.
Tying the Findings Into Theoretical Foundations
Researchers have proposed a number of theoretical perspectives that offer insights into why we are attracted to those whom we find similar to ourselves. Decades of work on the link between similarity and attraction have yielded a number of unique insights that may converge on several general explanations. Generally, we tend to like those with whom we enjoy interacting (Treger et al., 2013), those who may share a subjective experience with us (Pinel et al., 2006), and even those who found the same musical album as interesting as we did (Jones et al., 2004). Perhaps similarity in domains characterized by subjective experiences, such as in food or in music, are seen as important because they open the avenue to enjoy the time spent with another person, potentially validating their worldview (Byrne et al., 1973) and even allowing us to learn something new about the person (Kaplan & Anderson, 1973; Reis et al., 2011). This idea may be further evidenced by our discovery that political attitudes were the most important domain of similarity on our list. Political orientation affects a variety of facets of one’s worldviews and experiences, which may also include important norms for a particular lifestyle (e.g., liberals may select different regions of residence than may conservatives). Mismatches in such norms and styles of living may lead to conflict in a relationship, which in turn would drive dissolution. Indeed, the 2016 presidential election in the United States showed that people were readily willing to “unfriend” those on Facebook who share opposing political views (e.g., Lindner, 2016). Thus, perhaps people may foresee potential conflict stemming from such subjective experiences, which may demotivate them from starting a new relationship with another.
Further supporting these speculations are the domains that people rated as least important. The least important domains, such as cell phone brands and college majors, may further reflect theoretical insights from the decades of research on similarity and attraction. A college major, for example, may not necessarily be a driving force in affecting how much fun people can have with one another, nor how well they get along. Likewise, people of two different cell phone or computer brands may enjoy their time together just as much as two people who use the same brand. Perhaps subjective preference led them to gravitate towards one brand of phone or another, but, preference for one brand of ubiquitous tool over another does not reflect any deeper differences individuals may have in personality or philosophy of life that can be approximated by, for example, political inclinations and career goals.
Strengths, Limitations, and Future Directions
Our research entails a number of strengths. First, our study was the first to our knowledge to examine perceived similarity up close—rather than examining similarity in the abstract, we delved into specific domains that may potentially affect one’s attraction to another. In other words, although general similarity may be attractive, it is made up of certain domains that may “make or break” attraction to another individual —perhaps a lack of similarity in musical tastes may decrease attraction, even with similar political views and food preferences. Alternatively, similarity in particular clusters of domains may increase attraction, buffering any effect of dissimilarity in one or more others. These ideas, however, require further investigation in future research. Second, we further complemented research on preferences in a partner’s traits (e.g., Lippa, 2007; Sprecher & Regan, 2002) by examining not necessarily merely the preferences for various characteristics themselves (e.g., the degree to which one desires an intelligent partner), but rather, the degree to which one desires a match in a particular characteristic or preference. Thus, we were able to examine preferences for particular domains relative to the rater’s own preferences. Third, we examined whether such domains differed across several types of relationships, rather than focusing on one. Finally, our sample was fairly large, went beyond a college age range, and contained a diverse set of participants across multiple age groups.
This research, however, presents limitations as well. Domains of similarity go beyond 18, and some may potentially be more important than the importance persons placed in the domains we had selected. This limitation, however, opens an avenue for further research. One direction is to examine whether similarity domains display different levels of fluctuation in importance over time as a function of particular variables. Prior research has demonstrated that people’s preferences for some partner traits change with age (Buunk et al., 2002). Thus, there is potential that preferences for particular domains may fluctuate over time as a function of age or perhaps other major life events (e.g., childbirth). Furthermore, it is possible that certain domains of similarity may emerge only at different stages of life. For example, perhaps similarity in domains such as parenting style is only important for those who desire a child (a person with no intention of having kids may not consider this as a domain). Another interesting future direction is the examination of the effects of similarity in particular domains on liking during social interactions. For example, research has demonstrated that people’s desires may not too accurately predict attraction for a partner (Eastwick, Finkel, & Eagly, 2011). Thus, perhaps similarity in particular domains may appear important to persons broadly, but have little effects after face-to-face interactions.