You’re like me and I like you: Mediators of the similarity–liking link assessed before and after a getting-acquainted social interaction

Abstract

Several mediators of the similarity–liking link were examined in a laboratory study that involved previously unacquainted individuals interacting to become acquainted and that also included a preinteraction manipulation of bogus similarity. In addition to replicating the robust finding of a positive association between similarity and liking in both a bogus stranger paradigm and a social interaction paradigm, we tested the roles of five potential mediators of the similarity–liking association: consensual validation, cognitive evaluation, certainty of being liked, fun and enjoyment, and self-expansion opportunity. The results indicated certainty of being liked to be the strongest mediator prior to a live interaction and fun and enjoyment to be the strongest mediator postinteraction. Consensual validation was also found to be a strong mediator postinteraction.

Keywords

Attraction liking mediation similarity social interaction

Social scientists have identified a number of variables that affect initial attraction, or general liking, in the formation of relationships (for reviews, see Fehr, 2008; Finkel & Eastwick, 2015). One of the most frequently examined predictors of liking is similarity. Numerous studies have established that similarity leads to attraction in early relationship formation (e.g., Byrne, 1971; Condon & Crano, 1988; Montoya, Horton, & Kirchner, 2008; Morry, Kito, & Ortiz, 2011). In past research, there has been an important distinction between actual similarity and perceived similarity, although seldom are both considered in the same study. In addition, rarely have variables that might mediate (or explain) the similarity–liking link—such as consensual validation and certainty of being liked—been examined empirically. The present study considers both actual (manipulated) similarity and perceived similarity, with a focus on the variables that mediate, and therefore explain, the association between similarity and liking.

Actual versus perceived similarity

Actual similarity refers to the degree to which two people share attitudes, attributes, or other characteristics and has been operationalized in two major ways. First, Byrne (1961, 1971) introduced an experimental manipulation of actual similarity with the use of the phantom other technique, also referred to as the bogus stranger paradigm. This method involves experimentally manipulating the degree of similarity of a bogus target to the participant, who is then asked to evaluate that bogus target. Second, in research conducted with either already-acquainted pairs or getting-acquainted pairs, actual similarity has been operationalized by measuring both partners’ characteristics and then calculating the differences between their scores (e.g., Bleske-Rechek, Remiker, & Baker, 2009; Duck & Craig, 1978). Perceived similarity refers to the degree to which a person believes the other to be similar to himself or herself. It has generally been operationalized by asking participants how similar they are to a target other, who could be an actual friend or a partner (e.g., Sprecher, 2013) or a stranger about whom the participants had just received factual or bogus information (e.g., Hoyle, 1993).

The few studies that have included measures of both actual and perceived similarity have shown that perceived similarity has a stronger association with liking than does actual similarity (e.g., Condon & Crano, 1988; Sprecher, 2014; Tidwell, Eastwick, & Finkel, 2013). In a meta-analysis of 313 studies on similarity and attraction, Montoya, Horton, and Kirchner (2008) found that actual similarity had a strong effect in studies conducted with the bogus stranger paradigm, a moderate effect in studies involving a brief interaction between previously unacquainted participants, and no effect in studies with existing friends and romantic partners. Conversely, the effect of perceived similarity on attraction was found to be strong in all three types of studies. These results support the notion that perceived similarity is a stronger and more reliable predictor of attraction, and liking in general, than actual similarity.

Explaining the similarity–liking link

Researchers have proposed several explanations for the association between similarity and liking (e.g., Montoya & Horton, 2014; Montoya et al., 2008; Singh, Wegener, et al., 2017; Sprecher, Treger, Hilaire, Fisher, & Hatfield, 2013a). Below, we discuss five mediators that may explain the similarity–liking link.¹

Consensual validation

It has been argued that being with a similar other—particularly someone who is similar in attitudes—provides consensual validation, which then leads to positive affect and liking (Byrne & Clore, 1970; Hampton & Sprecher, 2017). Consensual validation contributes to the effectance motive, which was described as a basic need for a consistent, logical, and accurate interpretation of the world by Montoya and Horton (2004). More specifically, because people have no objective, external criteria to determine the correctness or superiority of many of their beliefs, they compare their beliefs with others (Festinger, 1957). Because it is more affectively rewarding to be validated in one’s thoughts and beliefs than invalidated, the comparison is often strategic and done with similar others. Furthermore, the positive reinforcement received from being consensually validated in turn leads to liking for the other. Recently, Singh, Wegener, et al. (2017) noted that although early social psychological literature acknowledged that consensual validation may be one explanation for why similarity leads to attraction, validation has not received enough attention as a potential mediator of the similarity–liking effect. Using a bogus stranger manipulation of similarity, Singh, Wegener, et al. (2017) found that validation (measured with adjective items such as “confirmed,” “assured,” and “validated”) mediated similarity’s effect on attraction. In a survey study that will be discussed further in a section below, consensual validation (measured by such items as “Because of who my partner is, I feel reassured of my own views of the world”) was also found to mediate the effect of similarity on liking (Sprecher et al., 2013a).

Cognitive evaluation

Another proposed mediator for the similarity–liking link is (positive) cognitive evaluation. Because people often assume additional information beyond what is already known about others (Kaplan & Anderson, 1973) and regard themselves and their attitudes as positive, people are likely to infer positive information about others who they believe share similarities to them (Ajzen, 1974). Montoya and Horton (2004) proposed that perceived similarity leads to the belief that the target has other positive attributes, which then leads to liking. To examine the role of cognitive evaluation as a mediator of the relationship between similarity and attraction, Montoya and Horton (2004) asked participants to provide an evaluation of a bogus target and indicate the degree to which they liked the target. They found that the evaluation mediated the similarity–attraction link, but only when the evaluation was measured prior to attraction, which may have made it more salient. More recently, Montoya and Horton (2014) proposed a two-dimensional model of attraction and argued that attraction for another is based on evaluations of two dimensions of the other. One of these dimensions is the ability of the other to help accomplish one’s own goals and needs. Inferred positive characteristics (or positive cognitive evaluation) convey information of this capacity that another will be able to help accomplish one’s goals and needs.

Certainty of being liked

The other dimension in Montoya and Horton’s (2014) parsimonious model of interpersonal attraction is the perceived evaluation that the other will be willing to facilitate his/her goals, which is reflected in the belief that the other will like them. People have an inherent need to belong and therefore seek situations and relationships where they feel liked and cared for (Baumeister & Leary, 1995). In support of this supposition, the knowledge of being liked—also referred to as reciprocal liking—has been found to be a strong predictor of attraction (Aronson, Wilson, & Akert, 2002; Aronson & Worchel, 1966; Finkel & Eastwick, 2015). Furthermore, people tend to assume that they will be liked more by similar others than dissimilar others (e.g., Aronson & Worchel, 1966; Greitemeyer, 2010). Condon and Crano (1988) studied the relationship between actual (manipulated) similarity and attraction, while controlling for inferred evaluation (expectation of being liked). The researchers found that the similarity–attraction link was much weaker when statistically controlling for the expectation of being liked, suggesting evidence that certainty of being liked plays a mediating role in the similarity effect.

Fun and enjoyable interactions

Another explanation often presented for the positive association between similarity and liking is that similarity leads to the expectation of enjoying interaction (or actually does lead to fun and enjoyable interactions, if there is interaction). People tend to like others with whom they enjoy interacting (e.g., Burleson, 1998). Furthermore, research has found that enjoyable interactions can contribute to the formation of new relationships (Felmlee, 1995) and liking in new relationships (Sprecher et al., 2013a). Relatedly, Treger, Sprecher, and Erber (2013) found, in a social interaction study, that enjoyment and laughter during get-acquainted interactions led to liking. Furthermore, similarity can enhance the enjoyment of interactions (e.g., Burleson, Kunkel, & Birch, 1994), suggesting that similarity contributes to enjoyment, which then leads to attraction. In their study, Burleson, Kunkel, and Birch (1994) found that similarity in attitudes, values, and beliefs about communication between relationship partners enhanced the enjoyment of partners’ interactions and thereby increased attraction to the partner.

Self-expansion opportunity

Another potential mediator of the similarity–liking link is perceived opportunities for self-expansion. According to self-expansion theory (Aron & Aron, 1986), people seek to increase their potential efficacy. To increase potential efficacy, they seek physical and social resources, perspectives, and knowledge (Aron, Aron, & Norman, 2001), which can be obtained in relationships with others. As such, developing relationships help to expand the self (Aron et al., 2001; Aron, Norman, Aron, McKenna, & Heyman, 2000; Aron, Paris, & Aron, 1995), making close relationships a key source of self-expansion.

Although it has been argued (e.g., Aron, Steele, Kashdan, & Perez, 2006) that a dissimilar other may offer more opportunities (than a similar other) for self-expansion through the new perspectives and knowledge that can be offered, similarity in some domains may present opportunities for self-expansion due to the increased self-other overlap that can emerge. In fact, Sprecher, Treger, Fisher, Hilaire, and Grzybowski (2015), in a study employing the social interaction paradigm, found perceived similarity to be positively associated with both the perception of self-expansion opportunities offered by the other and liking of the other. Additionally, and in relation to the current study, Sprecher et al. (2015) found that self-expansion opportunity mediated the relationship between perceived similarity and liking.

Research examining more than one mediator of the similarity–liking link

In sum, each of the above variables (among others) has been argued to be the reason that similarity leads to liking. Of the limited research that has examined mediators of the similarity–liking link, however, generally only one mediator has been examined in any single study (e.g., Condon & Crano, 1988; Montoya & Horton, 2004). There has been far less research that has examined simultaneously the degree to which several factors mediate the effects of similarity on liking.

In one exception, Sprecher, Treger, Hilaire, Fisher, and Hatfield (2013a) examined how consensual validation, certainty of being liked, fun and enjoyment, and self-expansion opportunity mediated the association between perceived similarity and attraction. In their correlational research conducted with individuals reporting about an existing relationship, Sprecher et al. (2013a) found that each of the mediating variables uniquely mediated the similarity–attraction link. In another exception using a bogus stranger experimental method, Singh et al. (2007; Study 2) examined three possible mediators of the similarity–attraction link: inferred attraction (i.e., certainty of being liked), cognitive evaluation (i.e., judgment of the target on other characteristics), and affect. The strongest mediator was found to be inferred attraction (see also Singh, Chen, & Wegener, 2014). More recently, Singh, Wegener, et al. (2017) compared consensual validation as a mediator of the similarity–attraction effect simultaneously with positive affect and trust and found each of these variables to have mediating roles.

The current study

As discussed above, multiple lines of research have examined the similarity–attraction effect. Individual studies have tested the effect of actual or perceived similarity on attraction, with some examining independent and concurrent mediators of this association. No study, however, has considered several mediators of the similarity–attraction link in the context of both the bogus stranger approach and the brief-interaction laboratory study and therefore whether the effects of these mediators depend on the type of similarity (i.e., actual vs. perceived). Such a pursuit was the goal of the current study.

Due to the nature of the sample (discussed below), we refer to liking and not (romantic) attraction. While we believe there are more variables involved in interpersonal liking than Montoya and Horton’s (2014) proposed two-dimensional model, we recognize that their findings regarding cognitive evaluation suggest that this variable likely plays a role in initial liking after a get-acquainted interaction. Each of the potential mediators discussed above—consensual validation, certainty of being liked, fun and enjoyment, self-expansion opportunity, and cognitive evaluation—was examined in the present study, in effort to replicate and expand on Sprecher et al.’s (2013a) findings. To thoroughly test these variables as potential mediators and determine whether their role depends on the type of similarity, they were tested in two contexts: the bogus stranger method (reactions to manipulated similarity information) and in a brief social interaction (reactions after an interaction with another). These two paradigms afforded us the ability to explore the independent and concurrent functions of variables mediating the similarity–liking link while considering both actual and perceived similarity. Moreover, these paradigms allowed us to examine perceived similarity in two contexts: (1) perceived similarity based on bogus information measured prior to the interaction and (2) perceived similarity based on partners’ impressions measured after the interaction.

We hypothesized that consensual validation, positive cognitive evaluation, certainty of being liked, fun and enjoyment, and self-expansion opportunity would each mediate the relationship between the preinteraction similarity manipulation (i.e., actual similarity) and liking of the bogus partner (H1) and would also mediate the relationship between perceived similarity and liking of the partner, both before interaction (H2) and after interaction (H3).

Method

Participants

The sample analyzed for this study consisted of 174 students (109 women) from the behavioral sciences at two medium-sized U.S. Midwestern Universities. The greater number of female participants was due to the imbalanced sex composition of the participant pool, particularly at the major university for data collection. Data were originally collected from 190 individuals; however, we eliminated five dyads in which the participants reported that they knew their partners prior to the interaction, two participants who were assigned to interact with a confederate when the scheduled partner did not arrive, two participants who reported they had figured out the similarity manipulation before debriefing, and two participants who had at least one half of their survey data missing. Participants represented several ethnic backgrounds with a majority being European American/Caucasian (N = 118). Ages of the participants ranged from 18 to 38 (M = 19.98, SD = 2.36). Participants were compensated with class research credit, per the policies of their professors.

Procedure

Participants were assigned to dyads based on mutual availability in their schedules. Because of the large number of female participants in the subject pools, the majority of pairs formed were female–female dyads (n = 39), with a smaller sample of male–female dyads (n = 31) and male–male dyads (n = 16). Data were also retained from two participants in male–male dyads, whose partners had missing data and were excluded from analyses.

Each member of the dyad arrived for the study at different rooms in a university building and was greeted by his or her own experimenter. The participants were first asked to complete a self-descriptive rating form, which they were told would be shared with their future interaction partner. While the experimenter was preparing the similarity manipulation (described below), the participant completed an online survey that asked for demographic and background information. Then, the participant received a rating sheet that was ostensibly completed by the other participant (located in another room), but that was actually completed by the experimenter. After viewing the rating sheet, participants completed a brief online survey that asked about their initial reactions to the bogus other.

The major task of the experimental session was the get-acquainted interaction. After the participants completed the preliminary procedure described above, the members of the dyad were placed in the same room and instructed to “get acquainted” for 15 min in a face-to-face, structured interaction (to be described below). Upon completion of the interaction, participants were once again separated and completed an online assessment of their reactions to the interaction and partner.

The self-descriptive rating form and the similarity manipulation

A self-assessment form similar to that used in Sprecher (2014) was used, in which participants described themselves on 17 dichotomous items. Some items asked about specific preferences (e.g., Reality show or Sitcom? Coffee or Tea?); other items asked about personality or general characteristics (e.g., Spender or Saver? Sloppy or Neat freak?). After the participant completed the rating form, the experimenter (out of sight of the participant) created a bogus version of the form so that it appeared to have been completed by the other participant. Depending on the condition to which the participant was randomly assigned, the experimenter prepared the bogus version so that the pair had 14 of the 17 items either similar or dissimilar. The items selected to not conform to the condition—the three similar responses in the dissimilarity condition and the three dissimilar responses in the similarity condition—were randomly determined for each participant.

The self-disclosure task

Like many prior social interaction studies (Sedikides, Campbell, Reeder, & Elliot, 1999; Sprecher, Treger, & Wondra, 2013b), the participants were given a list of self-disclosure topics to structure their get-acquainted interaction. The items were adapted from a combination of Aron, Melinat, Aron, Vallone, and Bator’s (1997) Closeness Generating Procedure and Sedikides, Campbell, Reeder, and Elliot’s (1999) Relationship Closeness Induction Task, with several additional questions added to ensure that the topics discussed were wide in scope and contemporary, interesting, and relevant to the participants as possible. Questions were divided into three sets that gradually increased in intimacy. Example questions included: “What would constitute a “perfect” day for you?” (Set 1), “What do you value most in a friendship?” (Set 2), and “If you were told you would die next week, what are three things on your bucket list that you would want to do or accomplish?” (Set 3). Each segment of the interaction lasted 5 min. Although the participants were left alone for the duration of the get-acquainted interaction, experimenters entered the room at the end of each segment to instruct participants to move on to the next set of questions or to end the interaction (after the last segment). When the dyad completed the final segment of the interaction, the participants were separated again to complete an online survey with reaction questions.

Preinteraction measures

After reviewing the bogus rating sheet of their interaction partner (but before meeting their partner), the participants responded to several questions as described below.

Perceived similarity

Participants were asked to report the degree to which they thought they were similar to their future partner with 2 items, each on a 7-point Likert scale (Sprecher, 2014): “How much do you think you have in common with the other person?” (nothing or almost nothing to a great deal), and “How similar do you think you and the other person are likely to be?” (not at all to a great deal). These items had good internal consistency (α = .89).

Mediating variables

To measure the mediating variables, we modified items from Sprecher et al. (2013a) and Montoya and Horton (2004). Each of the mediating variables was assessed with 3 items that were followed by a 5-point Likert scale (not at all true to extremely true). The 3 items for assessing consensual validation had low reliability (α = .42), but with the removal of one of the 3 items, the reliability for consensual validation increased to .61; therefore, we used only 2 items for analyses that include this preinteraction mediating variable (“My future interaction partner will probably support my attitudes and ideas” and “My future interaction partner will likely be ‘validating’—that is, they will help to convince me that I am correct in how I approach life”). Internal consistency was acceptable for the other mediating variables: cognitive evaluation, α = .73 (e.g., “My future interaction partner is probably well-respected.”); certainty of being liked, α = .67 (e.g., “I think my future interaction partner will like me.”); fun and enjoyment, α = .69 (e.g., “My future partner and I will probably laugh during our interaction period.”); and self-expansion opportunity, α = .77 (e.g., “Interacting with my future partner would likely open the door to new experiences.”).

Liking

Participants were also asked to complete an interpersonal attraction scale, which included items from prior research (e.g., Byrne & Wong, 1962; Montoya & Horton, 2004), although the items were phrased to focus on expectations for liking. This scale consisted of 9 items, each measured on a 7-point Likert scale (strongly disagree to strongly agree). Example items are “I would like to get to know this person better” and “I would probably dislike talking with my future interaction partner at a party” (reverse-coded). These scores also had acceptable internal consistency (α = .85).

Postinteraction measures

After the 15-min get-acquainted interaction, participants were separated and completed an online survey that assessed the same variables as referred to above, except generally with more items, as participants had more information on which to form opinions.

Perceived similarity

Perceived similarity was measured with 10 items on four specific domains (attitudes and values, interests and leisure activities, social skills/interaction styles, and background characteristics), with 2 items per specific domain (e.g., “My partner’s interaction style is similar to my own”). In addition, the two global items in the preinteraction survey were used to measure perceived similarity (e.g., “How much do you think you have in common with the other person?”). All of the items were assessed on a 7-point Likert scale (strongly disagree to strongly agree) and averaged to calculate a total perceived similarity scale that had good reliability (α = .92).

Mediating variables

After the interaction, participants were also asked to report the degree to which they experienced each of the mediating variables (consensual validation, cognitive evaluation, certainty of being liked, fun and enjoyment, and self-expansion opportunity) in their interaction. The same 15 items used to rate the bogus stranger prior to the interaction were included, but the wording was altered to make them more suitable for the current situation (e.g., changes in tense). An additional 2 items per mediating variable (10 items total) were added to assess the mediating variables after the interaction. Additional items were added because, after the interaction, participants would have more information on which to base their judgments. Like the items used to measure the mediating variables prior to the interaction, these items were adapted from Sprecher et al. (2013a) and Montoya and Horton (2004). All of the items were measured on a 5-point Likert scale (not at all true to extremely true). In general, reliability of the indices created from the items was higher after the interaction for the mediating variables: consensual validation (α = .74), cognitive evaluation (α = .83), certainty of being liked (α = .79), fun and enjoyment (α = .84), and self-expansion opportunity (α = .81).

Liking

Participants were also asked after the interaction to complete several questions designed to assess liking. This measure consisted of 15 items, modified from McCroskey and McCain’s (1974) Interpersonal Attraction Scale, Montoya and Horton’s (2004) Interpersonal Attraction Questionnaire, and measures included in Sprecher et al. (2013a) and Tidwell, Eastwick, and Finkel (2013). Each of the items was measured on a 7-point Likert scale (strongly disagree to strongly agree). Different items were selected for the postinteraction liking measure because the participants were responding based on an actual interaction (rather than predicting liking based on bogus information). Example items included “I think my interaction partner could be a friend of mine” and “My interaction partner and I had a real connection.” Scores on these items were reliable (α = .90).

Results

We begin by reporting several analyses to replicate the similarity effect in the two paradigms employed in this study: reactions to bogus information and reactions after a social interaction. Therefore, we look at the effects of both actual (i.e., manipulated) and perceived similarity. Then, we turn to analyses that examine the role of the mediating variables in explaining this link, again in both contexts.

Replicating the similarity–liking link

Method of analysis

Before testing our hypotheses, we first verified whether our dual paradigm method successfully replicated past findings of the similarity–liking association. In these analyses, we statistically controlled for any variance due to the dyad unit (because the dyadic data violate the traditional independence assumption). This was done using partial (first-order) correlations and random effects analyses of variance (ANOVAs). Random-effects ANOVAs are different from fixed ANOVAs, in that random-effects ANOVAs can account for both the individual and the dyad unit as “random factors” (whereas fixed-effects ANOVAs only examine individuals as random factors). This form of analysis helps to account for the variance due to the dyad unit by allowing researchers to specify the dyad unit as a random factor in addition to the person—which is always considered a random factor in fixed-effects models (for further explanation, see Kenny, Kashy, & Bolger, 1998).

Descriptive statistics

Descriptive statistics for the major variables, for the total sample and broken down into the two bogus stranger conditions, are presented in Table 1. The mean responses to the various items generally indicated that participants perceived a relatively high level of similarity, scored moderately high on the mediating variables, and generally liked their interaction partners. Partial correlations between pre- and postinteraction variables were conducted for the entire sample, controlling for the dyad unit (see Table 2). Ratings of perceived similarity pre- and postinteraction were weakly associated, r(171) = .11, p = .167, ratings of liking pre- and postinteraction were significantly associated, r(171) = .32, p < .001, and ratings of the five mediating variables pre- and postinteraction were all significantly associated, with partial correlations ranging from .33 to .45.²

Table 1.
Descriptive statistics for the major variables for total sample and by condition.

All Participants (N = 174) Similar Bogus (N = 83) Dissimilar Bogus (N = 91)

Variable M SD α M SD α M SD α

Perceived similarity (pre) 3.88 1.55 .89 5.17 .92 .69 2.70 .94 .74

Perceived similarity (post) 5.35 1.03 .92 5.46 .93 .92 5.25 1.10 .92

Liking (pre) 4.74 .81 .85 4.99 .78 .87 4.52 .78 .80

Liking (post) 5.67 .74 .90 5.61 .73 .91 5.73 .76 .90

Consensual validation (pre) 4.09 .91 .61 4.27 .90 .56 3.92 .90 .63

Consensual validation (post) 5.38 .74 .74 5.33 .71 .74 5.42 .77 .74

Cognitive evaluation (pre) 4.47 .87 .73 4.52 .93 .80 4.44 .83 .65

Cognitive evaluation (post) 5.60 .84 .83 5.55 .80 .83 5.64 .88 .83

Certainty of being liked (pre) 4.11 .96 .67 4.32 .94 .67 3.92 .94 .65

Certainty of being liked (post) 5.08 .88 .79 5.06 .85 .79 5.09 .92 .80

Fun and enjoyment (pre) 5.18 .87 .69 5.26 .89 .70 5.10 .86 .67

Fun and enjoyment (post) 5.87 .81 .84 5.78 .78 .86 5.94 .83 .82

Self-expansion opportunity (pre) 4.61 .93 .77 4.48 .83 .72 4.74 1.00 .80

Self-expansion opportunity (post) 5.22 .85 .81 5.13 .75 .78 5.29 .93 .83

Note. Variables followed by “(pre)” were measured prior to the interaction and were based on the bogus information presented about the partner. Variables followed by “(post)” were measured after the interaction and were based on the actual partner. All mediating variables for the bogus partner (prior to the interaction) were forecasted.

Table 2.
Partial correlations among key variables, controlling for the dyad unit.

(2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15)

(1) Similarity manipulation^a .80* .10 .29* −.08 .19* −.06 .05 −.05 .21 −.02 .09 −.10 −.14 −.10

(2) Perceived similarity (pre) — .11 .42* −.09 .32* .00 .21 .01 .32* −.02 .19 −.08 −.04 −.07

(3) Perceived similarity (post) — .28* .66* .14 .55* .12 .47* .29* .58* .22 .66* .19* .65*

(4) Liking (pre) — .32* .48* .41* .47* .28* .52* .28* .69* .37* .41* .25

(5) Liking (post) — .13 .67* .25 .64*** .19* .57* .26 .78* .33* .71*

(6) Consensual validation (pre) — .36* .32* .20 .43*** .18* .41*** .16* .36*** .15*

(7) Consensual validation (post) — .37* .69* .33* .62* .39* .65* .37* .66*

(8) Cognitive evaluation (pre) — .45* .30* .16* .38* .24 .42* .20

(9) Cognitive evaluation (post) — .23 .48* .30* .61* .26 .68*

(10) Certainty of being liked (pre) — .39* .44* .26** .18* .17*

(11) Certainty of being liked (post) — .27* .74* .17* .68*

(12) Fun and enjoyment (pre) — .38* .40* .26*

(13) Fun and enjoyment (post) — .29* .75*

(14) Self-expansion opportunity (pre) — .33***

(15) Self-expansion opportunity (post) —

Note. ^aCoding of similarity manipulation: 0 = dissimilarity condition; 1 = similarity condition.

p < .001; p < .01; p < .05.

Actual similarity

Supporting the validity of the manipulation, participants in the similarity condition reported significantly greater levels of perceived similarity with the bogus partner (M = 5.17, SD = .92) than did participants in the dissimilarity condition (M = 2.70, SD = .94), F(1, 88) = 181.67, p < .001, partial η² = .83. Furthermore, this preinteraction similarity manipulation had a significant effect on participant ratings of liking the partner, F(1, 88) = 10.79, p = .002, partial η² = .22. Participants who were assigned to receive the bogus similarity information reported greater levels of liking the partner (M = 4.99, SD = .78) than those who received dissimilarity information (M = 4.52, SD = .78). This indicates that the bogus partner similarity manipulation (i.e., actual similarity) successfully affected liking. To ensure that the preinteraction manipulation of similarity did not overreach or distort postinteraction ratings, we examined whether the manipulation affected liking after the interaction. The preinteraction similarity manipulation had no significant effect on reports of liking the interaction partner after the interaction, F(1, 88) = .67, p = .42, partial η² = .02.

Perceived similarity

Upon finding support for the association between actual similarity and liking, we tested whether perceived similarity (measured before the interaction) was positively associated with greater liking preinteraction, as well as whether perceived similarity (measured after the interaction) was associated with greater liking postinteraction. Perceived similarity to the bogus partner was positively associated with liking for the bogus partner prior to the interaction, r(171) = .42, p < .001; additionally, perceived similarity to the interaction partner was positively associated with liking for the discussion partner after the interaction, r(171) = .66, p < .001. Like the similarity manipulation, preinteraction perceptions of similarity did not appear to overreach or distort postinteraction reports of liking, r(171) = −.09, p = .231. Taken together, this information suggests that postinteraction ratings of liking are predicted by perceived similarity to the actual partner, rather than by perceived similarity to the bogus target prior to the interaction.

Ancillary analyses

In further examination of the partial correlations (Table 2), it is worth noting that preinteraction ratings of perceived similarity to the bogus partner were significantly correlated with all the mediating variables measured prior to the interaction (except for self-expansion opportunity). However, there were no significant correlations between preinteraction ratings of perceived similarity and any of the postinteraction mediating variables about the interaction partner. In other words, perceived similarity to the bogus partner may have influenced participants’ ratings of the mediating variables before the interaction, but it did not appear to affect any of these variables after the interaction. While the preinteraction measures of the mediating variables (about the bogus partner) may have held some degree of influence on postinteraction liking for the interaction partner (rs ranged from .13 to .33), stronger associations on liking of one’s interaction partner were observed among the mediating variables measured after the interaction (rs ranged from .57 to .78). While these results again highlight the null or dwindling effects of preinteraction impressions on postinteraction ratings, they also provide initial evidence of bogus stranger paradigms and in-person interactions being empirically distinct methods of examining the effects of similarity on liking.

Tests of mediation models

Upon establishing adequate replication of past results, hypotheses regarding the full mediation models were tested using Process Model 4 (Hayes, 2013). The fit of all mediation models was assessed with a k = 20,000 bootstrap sample, while also controlling for the dyad unit. Significance exists if there is no zero within the confidence interval.

Preinteraction: Actual similarity–liking

We first predicted that consensual validation, positive cognitive evaluation, certainty of being liked, fun and enjoyment, and self-expansion opportunity would each mediate the association between the preinteraction similarity manipulation and liking of the bogus partner (H1). As displayed in Table 3 and Figure 1, the indirect effect between the similarity manipulation on liking of the bogus partner, through each of the mediating variables, showed a significant, positive effect only for certainty of being liked; that is, those participants who received bogus information that was manipulated to reflect partner similarity were more certain of being liked, which positively affected their liking of the bogus partner. The indirect paths through consensual validation, positive cognitive evaluation, fun and enjoyment, and self-expansion opportunity were not significant.

Table 3.
Mediation analysis for all similarity–liking models.

Forecasted Mediator Effect of Manipulation on Mediator (a) Unique Effect of Mediator on Liking (b) Indirect Effect (ab) 95% CI

Lower Upper

Similarity manipulation and liking preinteraction; c: b = 0.47; c’: b = 0.33

Consensual validation .35* .08 .03 −.01 .09

Cognitive evaluation .08 .15* .01 −.02 .06

Certainty of being liked .41* .14* .06* .01 .14

Fun and enjoyment .16 .41* .07 −.04 .19

Self-expansion opportunity −.26 .11* −.02 −.10 .00

Perceived similarity and liking preinteraction; c: b = 0.22; c’: b = 0.13

Consensual validation .19* .05 .01 −.01 .04

Cognitive evaluation .12* .11* .01 .00 .03

Certainty of being liked .20* .13* .03* .01 .06

Fun and enjoyment .11* .41* .05* .01 .09

Self-expansion opportunity −.03 .12* .00 −.02 .01

Perceived similarity and liking postinteraction; c: b = 0.47; c’: b = 0.16

Consensual validation .39* .17* .07* .02 .13

Cognitive evaluation .39* .10 .03 .00 .09

Certainty of being liked .50* −.13* −.07 −.14 .00

Fun and enjoyment .52* .44* .23* .13 .34

Self-expansion opportunity .53* .12 .07 .00 .14

Note. The manipulation variable was coded as 1 (dissimilar bogus other) and 2 (similar bogus other). The total effect of the IV on the DV is represented as c path. The direct effect of the IV on the DV when controlling for the mediators is c’ path. All coefficients reported for paths a, b, and ab are unstandardized slopes. CIs of each indirect effect are based on 20,000 resamples.

p < .05.

Figure 1.
Mediation model for H1, illustrating mediation of the relationship between the similarity manipulation and attraction to the bogus partner (prior to the interaction). Unstandardized coefficients are listed with standard errors following in the parentheses. p < .05.

However, the similarity manipulation did appear to have a positive effect on reports of consensual validation; additionally, cognitive evaluation, fun and enjoyment, and self-expansion opportunity held significant positive effects on liking. Thus, our analysis suggests each of these variables to play a role in the similarity–liking link based on bogus information, though some variables may be more proximal to the bogus information (i.e., consensual validation and certainty of being liked), while others may be more distal and more closely related to liking (i.e., cognitive evaluation, fun and enjoyment, and self-expansion opportunity). As previously tested, the direct effect of the bogus partner similarity manipulation on liking of the bogus partner was significant (in the positive direction). While the β weight representing the direct effect between these two variables (β = .33) was less than the path coefficient of the total effect (β = .47), only partial mediation was evident (i.e., the direct effect remained significant; see Figure 1). Thus, H1 was partially supported.

Preinteraction: Perceived similarity (based on bogus information before interaction)-liking

Our second hypothesis was that consensual validation, positive cognitive evaluation, certainty of being liked, fun and enjoyment, and self-expansion opportunity would each mediate the relationship between the perceived similarity (prior to interaction) and liking of the bogus partner. Results were somewhat similar to the pattern of results from the first mediation model. As displayed in Table 3 and Figure 2, the indirect effect of perceived similarity and liking of the partner (based on bogus information) was again significant (in the positive direction) through certainty of being liked. In this second model, however, fun and enjoyment was found as an additional significant mediator. Again, consensual validation, positive cognitive evaluation, and self-expansion opportunity did not mediate the relationship between similarity and liking (although greater perceived similarity to the bogus partner did positively predict consensual validation, self-expansion opportunity did positively predict liking, and positive cognitive evaluation had significant a and b paths). Partial mediation was again evident, as the direct effect (while still significant) was less than the total effect, leaving the second hypothesis partially supported.

Figure 2.
Mediation model for H2, illustrating mediation of the relationship between perceived similarity to the bogus partner and attraction to the bogus partner (prior to the interaction). Unstandardized coefficients are listed with standard errors following in the parentheses. p < .05.

Postinteraction: Perceived similarity (to partner after interaction)-liking

Finally, we predicted that perceived similarity (after interaction) and liking of the partner (also after interaction) would be mediated by consensual validation, positive cognitive evaluation, certainty of being liked, fun and enjoyment, and self-expansion opportunity. A significant, positive indirect effect was again found for fun and enjoyment (see Table 3 and Figure 3). Additionally, consensual validation emerged as a significant mediator postinteraction. Despite its mediating role prior to interaction, certainty of being liked did not emerge as a significant mediator postinteraction; however, it is interesting to note its effect on liking was in the negative direction (an issue that will be discussed in exploratory analyses below). Once again, positive cognitive evaluation and self-expansion opportunity (while having significant a paths) did not function as successful mediators. Keeping in line with the past two models, the direct effect of perceived similarity on liking of the actual partner was less than the total effect. However, the direct effect remained significant, indicating partial mediation for our third model (thus, partially supporting H3).

Figure 3.
Mediation model for H3, illustrating mediation of the relationship between perceived similarity to the actual partner and attraction to the actual partner (after the interaction). Unstandardized coefficients are listed with standard errors following in the parentheses. p < .05.

Exploratory analyses

As stated above, in describing our third mediation model (perceived similarity–liking, postinteraction), certainty of being liked appeared to exert a negative effect on liking of the interaction partner. This finding may appear counterintuitive, given the positive, bivariate association between certainty of being liked and liking of the interaction partner (r = .57, p < .001). As such, we conducted further analyses to explore this effect. Given that certainty of being liked held strong, positive correlations with the other proposed mediators postinteraction, we first checked for any evidence of multicollinearity by assessing the variance inflation factors among the variables in our model. No strong evidence was found in support of this explanation (variance inflation factor ranged from 2.04 to 3.16, below a commonly recommended cutoff value of 5; Sheather, 2009).

Upon ruling out this explanation, we assessed whether any of the other proposed mediators in our model were suppressing the effect of certainty of being liked on liking. Given the large indirect effect of similarity on liking, through fun and enjoyment (b = .23), we tested whether removing this mediator from the model would significantly bolster the effect of certainty of being liked on liking. Indeed, upon removal of this variable, certainty of being liked held a positive (yet nonsignificant) effect on liking of the interaction partner (b = .01). Only when certainty of being liked was tested as a single mediator of the similarity–liking link was its b path and indirect effect significant. Moreover, the omission of fun and enjoyment increased the effects of each of the other proposed mediators on liking (consensual validation, b = .21; cognitive evaluation, b = .15; and self-expansion opportunity, b = .23), resulting in significant indirect effects for both cognitive evaluation (.06, CI [.01, .12]) and self-expansion opportunity (.13, CI [.05, .21]). Implications of these effects will be discussed below.

Discussion

The examination of the link between similarity (both actual and perceived) and liking is well established and has been consistently replicated in the past decades (e.g., Byrne, 1971; Condon & Crano, 1988; Montoya et al., 2008; Morry et al., 2011). Only a few studies, however, have examined mechanisms through which this association can be explained (e.g., Condon & Crano, 1988; Montoya & Horton, 2004). Even fewer studies have explored multiple explanations simultaneously (for an exception, see Sprecher et al., 2013a). The goal of this study was to replicate the similarity–liking association and to examine the possible role of five unique variables that may explain this link. To fully examine each of the proposed mediating variables (consensual validation, cognitive evaluation, certainty of being liked, fun and enjoyment, and self-expansion opportunity) and the role they play in the similarity–liking association, both a bogus stranger paradigm and a social interaction paradigm were employed. This allowed for the testing of the same model with two methods and allowed us to consider how the mediators functioned for both perceived similarity and actual similarity.

Replicating the similarity–liking link

The similarity–liking effect was replicated in three different ways. First testing the effect of actual similarity, we found that participants who received bogus similarity information about their future discussion partner reported higher levels of liking for that partner than participants who received bogus dissimilarity information about their partner. Second, we found that participants who perceived greater levels of similarity based on bogus information reported greater levels of liking before an interaction. Third, participants who perceived greater levels of similarity after interacting with a partner reported greater levels of liking for that partner. Similar to Montoya et al.’s (2008) findings, we found that actual similarity had a moderate effect on liking, whereas perceived similarity had stronger effects on liking. Finally, we found that reports of liking after the interaction were not biased due to previously received bogus information. This is consistent with Eastwick, Finkel, and Eagly (2011), who found that the effects of a profile manipulation were nonsignificant after a live interaction.

Mechanisms underlying the similarity–liking link

Our primary goal in this manuscript was to explore the effectiveness of five variables proposed to mediate the similarity–liking link. For variables to be considered successful mediators, they were required to meet several criteria. They needed to have a significant relation with the predictor (similarity), a significant relation with the outcome (liking), and a significant indirect effect (or amount of mediation). Below, we will discuss how each of the proposed mediating variables fared across our three models (actual similarity–liking before an interaction, perceived similarity–liking before an interaction, and perceived similarity–liking after an interaction).

Cognitive evaluation and self-expansion opportunity

Although there were some significant indirect paths in some of our models, generally cognitive evaluation and self-expansion opportunity did not serve as successful mediators. These variables are complex, and perhaps the limited 15-min structured interaction did not provide enough information on which participants could base these judgments. It may be that individuals need more time and information from others before being able to make these kinds of evaluations with confidence. The live interaction could explain why our results were different from Sprecher et al. (2013a). Because responses in the Sprecher et al.’s (2013a) study were based on an existing close other, there was much more information on which judgments could be based (even when recalling impressions from the initial acquaintance period).

Consensual validation

Consensual validation was only a significant mediator of the similarity–liking link after the interaction. It is likely that there was not enough information prior to the interaction for participants to feel validated. Alternatively, the kinds of items used to manipulate similarity (e.g., coffee or tea and radio or mp3 player) may not have been consequential enough descriptors for our participants to expect to feel personally validated in their worldview. As previously mentioned, there would have been much richer information after the interaction to facilitate participant reactions and judgments.

Certainty of being liked

In both of the models that examined liking of the partner based on bogus information prior to the interaction (i.e., effects of bogus information on similarity and perceived similarity based on bogus information), certainty of being liked successfully and positively mediated the similarity–liking link. These results should not be surprising, given people’s tendency to assume that they will be liked more by similar others than by dissimilar others (e.g., Aronson & Worchel, 1966; Greitemeyer, 2010). While certainty of being liked may exert some degree of influence on liking after a live interaction, this effect was suppressed by the other mediators (primarily fun and enjoyment) and only emerged as significant when tested as a single mediator. It may be that our participants did not expect to ever interact with their discussion partner again; as such, they may have placed little weight in what their partner thought of them when making judgments of their own liking. We do urge some degree of caution in interpreting these results, however, due to our college student sample. Upon entering college, many individuals are still concerned with finding individuals who like them as they are (Berzoff, 2015; Blisten, 1953).

Fun and enjoyment

Fun and enjoyment was perhaps the most intriguing mediator examined in the current study. Perceived similarity to the bogus partner, but not the similarity manipulation, exerted a positive influence on expected fun and enjoyment with a future interaction partner. While the disparity between these a paths in the first two models is unclear, it is evident by the large b paths that expected fun and enjoyment is an important contributor to how much one expects to like a future interaction partner.

More interesting, however, is the role that reported fun and enjoyment held in the similarity–liking link postinteraction. This variable was the strongest mediator of the similarity–liking link not only after participants got acquainted, but across all three of our models. As discussed in our exploratory analyses, fun and enjoyment suppressed the effects of the other proposed mediators on liking, and upon removal of fun and enjoyment from the model, their indirect effects were bolstered. These results suggest that as long as individuals have a fun and enjoyable time with another, little else matters; that is, they may place less weight on the other proposed mechanisms when evaluating the interaction and their partner (i.e., self-expansion opportunity, cognitive evaluation, and certainty of being liked). Again, there is also the possibility that our relatively young sample contributed to the effectiveness of this variable. College students might place more importance on fun and enjoyment over other qualities in forming new, interpersonal relationships. It would be interesting to see whether similar findings are obtained in an older sample, who might place more importance on other qualities in their relationships.

A two-dimensional model of general attraction

Taken together, our tests of the five proposed mediators of the similarity–liking link suggest a few general implications. While it appears that each of the proposed mediators plays some role in the similarity–liking link, our analyses suggest that individuals may place more weight on single rather than on multiple mediators when determining judgments of liking from similarity. For instance, when making judgments based on limited profile information (prior to meeting someone), feeling validated in one’s views or expecting opportunities for self-expansion likely come secondary to how certain one is of being liked. However, after getting to know someone better (i.e., postinteraction), how certain someone is on whether they were liked (among the other proposed mediators) may only matter to the extent that they had a fun and enjoyable experience (as discussed in our exploratory analyses). Together, these results suggest the potential for a two-dimensional model of general attraction with certainty of being liked holding weight when little is known about someone and fun and enjoyment holding weight after getting acquainted, a topic that should be explored further in future research.

Strengths, limitations, and future research

A primary strength of this study is the inclusion of five potential mediator variables in a single multiple mediation model that was tested several times, extending prior research in this area. Our goals for this study were to verify the original four mediators proposed in Sprecher et al. 2013a (consensual validation, certainty of being liked, level of fun and enjoyment, and self-expansion opportunity) and to build on the multiple mediation model by adding a fifth mediator (cognitive evaluation) that likely explains a unique part of the similarity–liking relationship. Additionally, we examined the proposed mediation model using two separate paradigms, each with its own strengths. The bogus stranger paradigm allowed us to examine causal relationships and look at the effect of actual similarity on liking and on expectations for the mechanisms (e.g., consensual validation) in the upcoming interaction. In addition, it allowed us to assess how several mechanisms mediated the effect of similarity on liking. The major advantage of the social interaction paradigm is that it provided the ecological validity that the bogus stranger paradigm lacked. Because we used both paradigms, we were able to examine how various factors mediate the similarity–liking association while considering both actual (manipulated) similarity and perceived similarity.

Despite the strengths of this research, there are also limitations that should be acknowledged. First, the sample size is relatively limited and homogenous due to the nature of a university sample. The large majority of participants were young adults studying the behavioral sciences. Additionally, because the participant pools, particularly at our major university for data collection, had a much larger number of females than that of males, we were not able to obtain as many male–male dyads, compared with female–female and female–male dyads. Because we found partial and not full mediation, it is also evident that there may be other mediating variables of the similarity–liking link that were not accounted for in this study. In hindsight, the exclusion of trust as a potential mediator is a limitation for this study. Another potential limitation lies in the divergent measures used from pre- to postinteraction. Because impressions were assessed in two ways (based on bogus information and based on interaction), different scales were selected to measure similarity, liking, and the proposed mediators. While many questions did overlap from pre- to postinteraction and each of the scales utilized has been previously validated, some caution may still be beneficial in interpreting these results. Finally, additional limitations lie in the correlational, rather than experimental, nature of the mediator data. There have been some recent studies reporting experimental measures or a combination of correlational and experimental measures (e.g., Singh, Tay, & Sankaran, 2017; Singh, Tengh, Bhullar, & Sakaran, 2017; Singh, Wegener, et al., 2017).

Perceived similarity assessed after the interaction was much stronger in predicting the mediating variables and liking than actual or perceived similarity based on the bogus information before the interaction. It would be interesting for future researchers to examine whether a similarity manipulation for topics of more significance (such as attitudes on controversial issues) would perhaps have a stronger effect on perceptions of a target. It is possible that the innocuousness of the items (e.g., Which do you prefer: Frozen Yogurt or Ice Cream? Reality Show or Sitcom?) contributed to the fact that the manipulation did not have stronger effects on the mediating variables and liking, particularly after interaction. Additionally, it would be helpful for future research to have longitudinal follow-ups with participants to see whether the pairs have future interactions. Similarity and/or any of the proposed mediating variables (e.g., experience of fun and enjoyment) could possibly be used to predict which dyads would choose to see each other again. Perhaps future research, with even more data points, can advance the field even further by determining causal orders of the mediators.

Conclusion

Regardless of one’s age or sex, similarity is attractive. Considerable research examining the association between similarity and liking has found a consistently strong link. Many mechanisms could explain this relationship, and this study attempted to examine the roles of five of these mechanisms in three mediation models. We identified two variables as particularly successful mediators of the similarity–liking link: certainty of being liked (prior to an interaction) and fun and enjoyment (postinteraction). In other words, similarity may be more attractive because we expect to like anonymous others who will likely view us positively and because we experience higher levels of fun and enjoyment with similar others during an interaction. It is important to continue research to understand the relationship between similarity and liking, as each is an essential aspect in human relationships.

	All Participants (N = 174)	Similar Bogus (N = 83)	Dissimilar Bogus (N = 91)
Perceived similarity (pre)	3.88	1.55	.89	5.17	.92	.69	2.70	.94	.74
Perceived similarity (post)	5.35	1.03	.92	5.46	.93	.92	5.25	1.10	.92
Liking (pre)	4.74	.81	.85	4.99	.78	.87	4.52	.78	.80
Liking (post)	5.67	.74	.90	5.61	.73	.91	5.73	.76	.90
Consensual validation (pre)	4.09	.91	.61	4.27	.90	.56	3.92	.90	.63
Consensual validation (post)	5.38	.74	.74	5.33	.71	.74	5.42	.77	.74
Cognitive evaluation (pre)	4.47	.87	.73	4.52	.93	.80	4.44	.83	.65
Cognitive evaluation (post)	5.60	.84	.83	5.55	.80	.83	5.64	.88	.83
Certainty of being liked (pre)	4.11	.96	.67	4.32	.94	.67	3.92	.94	.65
Certainty of being liked (post)	5.08	.88	.79	5.06	.85	.79	5.09	.92	.80
Fun and enjoyment (pre)	5.18	.87	.69	5.26	.89	.70	5.10	.86	.67
Fun and enjoyment (post)	5.87	.81	.84	5.78	.78	.86	5.94	.83	.82
Self-expansion opportunity (pre)	4.61	.93	.77	4.48	.83	.72	4.74	1.00	.80
Self-expansion opportunity (post)	5.22	.85	.81	5.13	.75	.78	5.29	.93	.83

	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)	(13)	(14)	(15)
(1) Similarity manipulation^a	.80***	.10	.29***	−.08	.19*	−.06	.05	−.05	.21**	−.02	.09	−.10	−.14	−.10
(2) Perceived similarity (pre)	—	.11	.42***	−.09	.32***	.00	.21**	.01	.32***	−.02	.19**	−.08	−.04	−.07
(3) Perceived similarity (post)		—	.28***	.66***	.14	.55***	.12	.47***	.29***	.58***	.22**	.66***	.19*	.65***
(4) Liking (pre)			—	.32***	.48***	.41***	.47***	.28***	.52***	.28***	.69***	.37***	.41***	.25**
(5) Liking (post)				—	.13	.67***	.25**	.64***	.19*	.57***	.26**	.78***	.33***	.71***
(6) Consensual validation (pre)					—	.36***	.32***	.20**	.43***	.18*	.41***	.16*	.36***	.15*
(7) Consensual validation (post)						—	.37***	.69***	.33***	.62***	.39***	.65***	.37***	.66***
(8) Cognitive evaluation (pre)							—	.45***	.30***	.16*	.38***	.24**	.42***	.20**
(9) Cognitive evaluation (post)								—	.23**	.48***	.30***	.61***	.26**	.68***
(10) Certainty of being liked (pre)									—	.39***	.44***	.26**	.18*	.17*
(11) Certainty of being liked (post)										—	.27***	.74***	.17*	.68***
(12) Fun and enjoyment (pre)											—	.38***	.40***	.26***
(13) Fun and enjoyment (post)												—	.29***	.75***
(14) Self-expansion opportunity (pre)													—	.33***
(15) Self-expansion opportunity (post)														—

Forecasted Mediator	Effect of Manipulation on Mediator (a)	Unique Effect of Mediator on Liking (b)	Indirect Effect (ab)	95% CI
Similarity manipulation and liking preinteraction; c: b = 0.47; c’: b = 0.33
Consensual validation	.35*	.08	.03	−.01	.09
Cognitive evaluation	.08	.15*	.01	−.02	.06
Certainty of being liked	.41*	.14*	.06*	.01	.14
Fun and enjoyment	.16	.41*	.07	−.04	.19
Self-expansion opportunity	−.26	.11*	−.02	−.10	.00
Perceived similarity and liking preinteraction; c: b = 0.22; c’: b = 0.13
Consensual validation	.19*	.05	.01	−.01	.04
Cognitive evaluation	.12*	.11*	.01	.00	.03
Certainty of being liked	.20*	.13*	.03*	.01	.06
Fun and enjoyment	.11*	.41*	.05*	.01	.09
Self-expansion opportunity	−.03	.12*	.00	−.02	.01
Perceived similarity and liking postinteraction; c: b = 0.47; c’: b = 0.16
Consensual validation	.39*	.17*	.07*	.02	.13
Cognitive evaluation	.39*	.10	.03	.00	.09
Certainty of being liked	.50*	−.13*	−.07	−.14	.00
Fun and enjoyment	.52*	.44*	.23*	.13	.34
Self-expansion opportunity	.53*	.12	.07	.00	.14

Footnotes

Acknowledgements

The authors would like to thank all of the following research assistants who helped with data collection: Andrew Clesen, Caitlin Hannigan, Madeline Luniewicz, Michelle Malik, Jennifer Sandoval, Chaunte Wilder, and Hayley Green. In addition, the authors would like to thank W. Joel Schneider, Eric Wesselmann, and J. Cooper Cutting for comments on an earlier version of the manuscript. For correspondence on this manuscript, contact the first author at hampto13@purdue.edu or the third author at sprecher@ilstu.edu.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Open research statement

As part of International Association for Relationships Research’s encouragement of open research practices, the authors have provided the following information: This research was not preregistered. The data used in the research are not available. Materials can be made available by writing to the first author, Adam J. Hampton, at hampto13@purdue.edu.

Notes

References

Ajzen

(1974). Effects of information on interpersonal attraction: Similarity versus affective value. Journal of Personality and Social Psychology, 29, 374–380. doi:10.1037/h0036002

Aron

E. N.

(1986). Love and the expansion of self: Understanding attraction and satisfaction. New York, NY: Hemisphere.

Aron

E. N.

Norman

(2001). Self-expansion model of motivation and cognition in close relationships and beyond. In Clark

Fletcher

(Eds.), Blackwell’s handbook of social psychology, Vol. 2: Interpersonal processes (pp. 478–502). Oxford, England: Blackwell.

Aron

Melinat

Aron

E. N.

Vallone

Bator

(1997). The experimental generation of interpersonal closeness: A procedure and some preliminary findings. Personality and Social Psychology Bulletin, 23, 363–377. doi:10.1177/0146167297234003

Aron

Norman

C. C.

Aron

E. N.

McKenna

Heyman

R. E.

(2000). Couples’ shared participation in novel and arousing activities and experienced relationship quality. Journal of Personality and Social Psychology, 78, 273–284. doi:10.1037/0022-3514.78.2.273

Aron

Paris

Aron

E. N.

(1995). Falling in love: Prospective studies of self-concept change. Journal of Personality and Social Psychology, 69, 1102–1112. doi:10.1037/0022-3514.69.6.1102

Aron

Steele

J. L.

Kashdan

T. B.

Perez

(2006). When similars do not attract: Tests of a prediction from the self-expansion model. Personal Relationships, 13, 387–396. doi:10.1111/j.1475-6811.2006.00125.x

Aronson

Wilson

T. D.

Akert

R. M.

(2010). Interpersonal attraction: From first impressions to close relationships. In Baron

R. A.

Branscombe

N. R.

(Eds.), Social psychology (13th ed., pp. 286–321). Upper Saddle River, NJ: Prentice Hall.

Aronson

Worchel

(1966). Similarity vs. liking as determinants of interpersonal attractiveness. Psychonomic Science, 5, 157–158. doi:10.3758/BF03328329

10.

Baumeister

R. F.

Leary

M. R.

(1995). The need to belong: Desire for interpersonal attachments as a fundamental human motivation. Psychological Bulletin, 117, 497–529. doi:10.1037/0033-2909.117.3.497

11.

Berzoff

(2015). H. S. Sullivan and his contribution to the relational school. Smith College Studies in Social Work, 85, 54–71.

12.

Bleske-Rechek

Remiker

M. W.

Baker

J. P.

(2009). Similar from the start: Assortment in young adult dating couples and its link to relationship stability over time. Individual Differences Research, 7, 142–158.

13.

Blisten

D. R.

(1953). The social theories of Harry Stack Sullivan. New York, NY: Academic Press.

14.

Burleson

B. R.

(1998). Similarities in social skills, interpersonal attraction, and the development of personal relationships. In Trent

J. S.

(Ed.), Communication: Views from the helm for the 21st century (pp. 77–84). Boston, MA: Allyn and Beacon.

15.

Burleson

B. R.

Kunkel

A. W.

Birch

J. D.

(1994). Thoughts about talk in romantic relationships: Similarity makes for attraction (and happiness, too). Communication Quarterly, 42, 259–273. doi:10.1080/01463379409369933

16.

Byrne

D. E.

(1961). Interpersonal attraction and attitude similarity. Journal of Abnormal and Social Psychology, 62, 713–715. doi:10.1037/h0044721

17.

Byrne

D. E.

(1971). The attraction paradigm. New York, NY: Academic Press.

18.

Byrne

D. E.

Clore

G. L.

(1970). A reinforcement model of evaluative responses. Personality: An International Journal, 1, 103–128.

19.

Byrne

D. E.

Wong

T. J.

(1962). Racial prejudice, interpersonal attraction, and assumed dissimilarity of attitudes. Journal of Abnormal and Social Psychology, 65, 246–253. doi:10.1037/h0047299

20.

Condon

J. W.

Crano

W. D.

(1988). Inferred evaluation and the relation between attitude similarity and interpersonal attraction. Journal of Personality and Social Psychology, 54, 789–797. doi:10.1037/0022-3514.54.5.789

21.

Duck

S. W.

Craig

(1978). Personality similarity and the development of friendship: A longitudinal study. British Journal of Social and Clinical Psychology, 17, 237–242. doi:10.1111/j.2044-8260.1978.tb00272.x

22.

Eastwick

P. W.

Finkel

E. J.

Eagly

A. H.

(2011). When and why do ideal partner preferences affect the process of initiating and maintaining romantic relationships? Journal of Personality and Social Psychology, 101, 1012–1032. doi:10.1037/a0024062

23.

Fehr

(2008). Friendship formation. In Sprecher

Wenzel

Harvey

J. H.

(Eds.), Handbook of relationship initiation (pp. 29–54). New York, NY: Taylor & Francis.

24.

Felmlee

D. H.

(1995). Fatal attractions: Affection and disaffection in intimate relationships. Journal of Social and Personal Relationships, 12, 295–311. doi:10.1177/0265407595122009

25.

Festinger

L. A.

(1957). A theory of cognitive dissonance. Evanston, IL: Row Peterson.

26.

Finkel

E. J.

Eastwick

P. W.

(2015). Interpersonal attraction: In search of a theoretical Rosetta Stone. In Simpson

J. A.

Dovidio

J. F.

(Eds.), APA handbook of personality and social psychology, Vol. 3: Interpersonal relations (pp. 179–210). Washington, DC: American Psychological Association.

27.

Greitemeyer

(2010). Effects of reciprocity on attraction: The role of a partner’s physical attractiveness. Personal Relationships, 17, 317–330. doi:10.1111/j.1475-6811.2010.01278.x

28.

Hampton

A. J.

Sprecher

(2017). Consensual validation. In Ziegler-Hill

Shackelford

(Eds.), Encyclopedia of personality and individual differences. Advance online publication. New York, NY: Springer. doi:10.1007/978-3-319-28099-8_1287 -1

29.

Hayes

A. F.

(2013). Introduction to mediation, moderation, and conditional process analysis. New York, NY: The Guilford Press.

30.

Hoyle

R. H.

(1993). Interpersonal attraction in the absence of explicit attitudinal information. Social Cognition, 11, 309–320. doi:10.1521/soco.1993.11.3.309

31.

Kaplan

M. F.

Anderson

N. H.

(1973). Information integration theory and reinforcement theory as approaches to interpersonal attraction. Journal of Personality and Social Psychology, 28, 301–312. doi:10.1037/h0035112

32.

Kenny

D. A.

Kashy

Bolger

(1998). Data analysis in social psychology. In Gilbert

Fiske

Lindzey

(Eds.), Handbook of social psychology (4th ed., pp. 233–265). New York, NY: McGraw-Hill.

33.

McCroskey

J. C.

McCain

T. A.

(1974). The measurement of interpersonal attraction. Speech Monographs, 41, 261–266. doi:10.1080/03637757409375845

34.

Montoya

R. M.

Horton

R. S.

(2004). On the importance of cognitive evaluation as a determinant of interpersonal attraction. Journal of Personality and Social Psychology, 86, 696–712. doi:10.1037/0022-3514.86.5.696

35.

Montoya

R. M.

Horton

R. S.

(2014). A two-dimensional model for the study of interpersonal attraction. Personality and Social Psychology Review, 18, 59–86. doi:10.1177/1088868313501887

36.

Montoya

R. M.

Horton

R. S.

Kirchner

(2008). Is actual similarity necessary for attraction? A meta-analysis of actual and perceived similarity. Journal of Social and Personal Relationships, 25, 889–923. doi:10.1177/0265407508096700

37.

Morry

M. M.

Kito

Ortiz

(2011). The attraction-similarity model and dating couples: Projection, perceived similarity, and psychological benefits. Personal Relationships, 18, 125–143. doi:10.1111/j.1475-6811.2010.01293.x

38.

Sedikides

Campbell

W. K.

Reeder

G. D.

Elliot

A. J.

(1999). The relationship closeness induction task. Representative Research in Social Psychology, 23, 1–4.

39.

Sheather

S. J.

(2009). A modern approach to regression with R. New York, NY: Springer.

40.

Singh

Chen

Wegener

D. T.

(2014). The similarity-attraction link: Sequential versus parallel multiple-mediator models involving inferred attraction, respect, and positive affect. Basic and Applied Social Psychology, 36, 281–298.

41.

Singh

Tay

Y. Y.

Sankaran

(2017). Causal role of trust in interpersonal attraction from attitude similarity. Journal of Social and Personal Relationships, 34, 717–731. doi:10.1177/0265407516656826

42.

Singh

Teng

J. S. Q.

Bhullar

Sankaran

(2018). Positive affect as a moderator of liking effects on trust and attraction. Asian Journal of Social Psychology, 21, 13–21. doi:10.1111/ajsp.12192

43.

Singh

Wegener

D. T.

Sankaran

Bhullar

Ang

K. Q. P.

Chia

P. J. L.

… Chen

(2017). Attitude similarity and attraction: Validation, positive affect, and trust as sequential mediators. Personal Relationships, 24, 203–222. doi:10.1111/pere.12178

44.

Singh

Yeo

S. E.

Lin

P. K. F.

Tan

(2007). Multiple mediators of the attitude similarity-attraction relationship: Dominance of inferred attraction and subtlety of affect. Basic and Applied Social Psychology, 29, 61–74. doi:10.1080/01973530701331007

45.

Sprecher

(2013). Correlates of couples’ perceived similarity at the initiation stage and currently. Interpersona, 7, 180–195. doi:10.5964/ijpr.v7i2.126

46.

Sprecher

(2014). Effects of actual (manipulated) and perceived similarity on liking in get-acquainted interactions: The role of communication. Communication Monographs, 81, 4–27. doi:10.1080/03637751.2013.839884

47.

Sprecher

Treger

Fisher

Hilaire

Grzybowski

(2015). Associations between self-expansion and actual and perceived (dis)similarity and their joint effects on attraction in initial interactions. Self and Identity, 14, 369–389. doi:10.1080/15298868.2014.1003592

48.

Sprecher

Treger

Hilaire

Fisher

Hatfield

(2013a). You validate me, you like me, you’re fun, you expand me: “I’m yours!” Current Research in Social Psychology, 21, 22–34.

49.

Sprecher

Treger

Wondra

J. D.

(2013b). Effects of self-disclosure role on liking closeness, and other impressions in get-acquainted interactions. Journal of Social and Personal Relationships, 30, 497–514. doi:10.1177/0265407512459033

50.

Tidwell

N. D.

Eastwick

P. W.

Finkel

E. J.

(2013). Perceived, not actual, similarity predicts initial attraction in a live romantic context: Evidence from the speed-dating paradigm. Personal Relationships, 20, 199–215. doi:10.1111/j.1475-6811.2012.01405.x

51.

Treger

Sprecher

Erber

(2013). Laughing and liking: Exploring the interpersonal effects of humor use in initial social interactions. European Journal of Social Psychology, 43, 532–543. doi:10.1002/ejsp.196

	All Participants (N = 174)			Similar Bogus (N = 83)			Dissimilar Bogus (N = 91)
Variable	M	SD	α	M	SD	α	M	SD	α
Perceived similarity (pre)	3.88	1.55	.89	5.17	.92	.69	2.70	.94	.74
Perceived similarity (post)	5.35	1.03	.92	5.46	.93	.92	5.25	1.10	.92
Liking (pre)	4.74	.81	.85	4.99	.78	.87	4.52	.78	.80
Liking (post)	5.67	.74	.90	5.61	.73	.91	5.73	.76	.90
Consensual validation (pre)	4.09	.91	.61	4.27	.90	.56	3.92	.90	.63
Consensual validation (post)	5.38	.74	.74	5.33	.71	.74	5.42	.77	.74
Cognitive evaluation (pre)	4.47	.87	.73	4.52	.93	.80	4.44	.83	.65
Cognitive evaluation (post)	5.60	.84	.83	5.55	.80	.83	5.64	.88	.83
Certainty of being liked (pre)	4.11	.96	.67	4.32	.94	.67	3.92	.94	.65
Certainty of being liked (post)	5.08	.88	.79	5.06	.85	.79	5.09	.92	.80
Fun and enjoyment (pre)	5.18	.87	.69	5.26	.89	.70	5.10	.86	.67
Fun and enjoyment (post)	5.87	.81	.84	5.78	.78	.86	5.94	.83	.82
Self-expansion opportunity (pre)	4.61	.93	.77	4.48	.83	.72	4.74	1.00	.80
Self-expansion opportunity (post)	5.22	.85	.81	5.13	.75	.78	5.29	.93	.83