Introduction

One of the most important developmental tasks in young adulthood is to form a positive and stable romantic relationship (Arnett 2000; Havighurst 1972; Nikitin and Freund 2008; Roisman et al. 2004). Not surprisingly, then, young adults are highly motivated to initiate romantic relationships (Buss 1994; Wrosch and Heckhausen 1999) and romantic relationships are among the most significant relationships in young adulthood (Antonucci et al. 2010; Takahashi 2005).

There exists a large body of research on the initiation of romantic relationships (Sprecher 2008). The majority of this literature addresses strategies of relationship initiation (e.g., Clark et al. 1999; Eastwick et al. 2007), the role of partners’ similarity (e.g., Frazier et al. 1996), and general theories of relationship initiation such as attachment theory or evolutionary theories (e.g., Creasey and Jarvis 2008; Latty-Mann and Davis 1996; Schmitt 2008). Although social motives (i.e., what people generally desire for and fear of when entering a social situation) might be a powerful predictor of the decision to initiate social relationships in general, and romantic relationships in particular (Eiser and Fazio 2008; Nurmi et al. 1996), motivational research has largely neglected the role of social motives for relationship initiation.

Indirect evidence on the role of social motives comes from studies on self-esteem (Cameron et al. 2010, 2013; Stinson et al. 2012): Typically, high (compared to low) self-esteem leads to initiation of romantic relationships, particularly in situations that are ambivalent with regard to acceptance and rejection (e.g., Cameron et al. 2010; Stinson et al. 2015). This is the case because self-esteem (defined as a global self-evaluation; e.g., Leary and MacDonald 2003) determines whether an ambivalent social cue is interpreted as a potential opportunity for acceptance or as a potential threat of rejection. For example, Cameron et al. (2010) demonstrated in a series of studies that persons with lower levels of self-esteem underestimated and persons with higher levels of self-esteem overestimated acceptance by potential romantic partners in ambivalent social situations. Cameron et al. argued that this is the case because persons with lower levels of self-esteem are motivated to avoid rejection and, thus, protect themselves from rejection by underestimating acceptance by potential romantic partners, whereas persons with higher levels of self-esteem are motivated to promote new relationships and, thus, overestimate acceptance.

Supporting the notion that self-esteem is differentially linked to social approach and avoidance motivation, previous research found that high self-esteem is associated with relatively higher levels of social approach motivation (i.e., the motivation to approach social rewards) compared to social avoidance motivation (i.e., the motivation to avoid social threats), whereas low self-esteem is associated with relatively lower levels of social approach compared to avoidance motivation (Heimpel et al. 2006). Consequently, social approach motivation should facilitate and social avoidance motivation should hinder the initiation of romantic relationships. The present research tests this hypothesis. In addition, the present research investigates possible mechanisms that underlie the differential association of social approach and avoidance motivation with relationship initiation. Social approach and social avoidance motivation are conceptualized as two independent motivational orientations that are incentivized by different outcomes: Whereas social approach motivation is driven by the presence or absence of positive social outcomes, social avoidance motivation is driven by the presence or absence of negative social outcomes (e.g., Gable and Berkman 2008; Nikitin and Schoch 2014; see also Higgins 1998).

In the current studies, we hypothesize that social approach motives (i.e., the dispositional motivation to approach positive social outcomes) and social avoidance motives (i.e., the dispositional motivation to avoid negative social outcomes; Gable and Berkman 2008; Mehrabian 1994) predict relationship initiation through different mechanisms. We propose that social approach motives predict relationship initiation because they are associated with expectations of a successful social interaction based on attributions of previous positive experiences with similar situations. In contrast, we posit that social avoidance motives predict relationship initiation because they are related to expectations of a negative social interaction based on attributions of previous negative experiences with similar situations. In the current studies, we investigated how perceived social rewards and social threats translate into relationship initiation dependent on people’s social approach and avoidance motives. In doing so, the present research contributes to understanding the motivational underpinnings of relationship initiation and their underlying cognitive mechanisms (i.e., attributions and expectations).

We tested our research questions using the sample case of speed dating, as the speed-dating paradigm seems to be particularly well-suited for the study of relationship initiation (Eastwick and Finkel 2008). Study 1 tested whether social approach and social avoidance motives predict the decision to participate in a speed-dating event. Study 2 focused on mechanisms underlying the motive-related decision: attributions and expectations.

The role of social approach and avoidance motives for relationship initiation

The initiation of a romantic relationship involves complex decision making (see review by Joel et al. 2013) requiring the weighing of the benefits (the possibility to be accepted and loved) and the costs of the decision (the possibility to be rejected and hurt; McClure et al. 2010; Murray et al. 2006).

People differ in their relative focus on possible costs and benefits of a relationship initiation based on their social motives. Social approach motives are generally associated with a focus on positive cues in social situations. For example, Gable and Poore (2008) demonstrated that when people evaluate their relationships, persons high in social approach motives weigh positive features of their relationship more heavily than individuals low in approach motives. In contrast, social avoidance motives were not associated with a focus on positive features of the relationship. Similarly, Nikitin et al. (2012) found in a diary study that social approach (but not avoidance) motives predicted a positive perception of everyday social interactions. Thus, a person high in social approach motives should be more attracted by possible benefits (affiliation and love) in his or her decision to initiate a romantic relationship than a person low in social approach motives. Consequently, social approach motives should be positively associated with the decision to enter a situation that entails the possibility to initiate a romantic relationship. In line with this hypothesis, previous research found that the expectation of positive outcomes is a powerful predictor of the decision to get in touch with other people (Anthony et al. 2007; Thibaut and Kelley 1959). Moreover, people who attended a speed-dating event in a study by McClure and colleagues (McClure et al. 2010) were lower on insecure-avoidant attachment style (which is negatively associated with social approach motives, see Nikitin and Freund 2010, Study 1) than people who did not attend a speed-dating event.

In contrast to social approach motives, social avoidance motives are associated with a focus on negative social cues. In the study by Gable and Poore (2008), participants high in social avoidance motives weighed negative features of their relationships more heavily than persons low in social avoidance motives. In contrast, social approach motives were not associated with a focus on negative features of the relationship. Similarly, social avoidance (but not approach) motives predicted negative cognitions about daily social interactions (Nikitin et al. 2012) and negative outcomes in social events, e.g., being rejected in speed-dating events (McClure et al. 2010). Thus, a person high in social avoidance motives should be more concerned with possible costs such as rejection and pain in his or her decision to initiate a romantic relationship than a person low in social avoidance motives. Consequently, social avoidance motives should be negatively associated with the decision to initiate a romantic relationship. In line with this hypothesis, people who have negative expectations for an outcome (e.g., who think about possible failure) avoid risk decisions (Leder et al. 2013; Stinson et al. 2014). As the initiation of a romantic relationship bears the risk of rejection (e.g., MacDonald and Leary 2005; Murray et al. 2006), people high in social avoidance motives should decide against it more often than people low in social avoidance motives.Footnote 1

Study 1 tested these hypotheses. Study 2 focused on mechanisms underlying the differential prediction of social approach and avoidance motives for the initialization of a romantic relationship: attributions and expectations.

Attributions and expectations as underlying mechanisms

Expectations for future social situations are heavily influenced by previous experiences in similar situations (e.g., Eccles and Wigfield 2002). This association is moderated by the attribution of the causes of the previous social experiences (Anderson and Jennings 1980; Brodt and Zimbardo 1981; Sedikides and Alicke 2012). The central dimensions of causal attributions are internality, stability, and globality (Heider 1958; Weiner et al. 1987). Internality refers to whether the cause of an outcome is located within the person (internal) or in the situation or other persons (external). Stability refers to whether the cause of an outcome is relatively temporary (variable) or permanent (stable). Globality refers to whether the cause of an outcome is specific to a given situation (specific) or holds across various situations (global). The attribution of positive events to internal (vs. external), stable (vs. variable), and global (vs. specific) causes and the attribution of negative events to external (vs. internal), variable (vs. stable), and specific (vs. global) causes are associated with positive expectations for similar future situations (Mezulis et al. 2004). This is the case because attributing positive events to internal, stable, and global causes (such as ability or personality) fosters positive views of oneself (Taylor and Brown 1988). Such a positive self-concept, in turn, increases people’s confidence that they will master similar situations in the future. In contrast, attributing negative events to internal, stable, and global causes (such as lack of ability or personality flaws) leads to pessimism and hopelessness that are associated with the expectation for negative events are likely to recur in the future across a wide variety of domains (Abramson et al. 1989).

Why do some people attribute positive social outcomes (such as being liked) to internal, stable, and global causes while others are prone to attribute negative social outcomes (such as being rejected) in this manner? Based on the literature on social motives, we posit that social approach and avoidance motives are important for the attributions of positive and negative social outcomes. As discussed above, social approach motives are associated with a focus on potential positive outcomes. When these positive outcomes actually occur, people are more likely to attribute them to internal (vs. external), stable (vs. variable), and global (vs. specific) causes (Alden 1986). This is the case because causal attributions operate as a self-verification process. They help to maintain a consistent sense of self and the world (Alden 1986). The general view associated with social approach motives is that social situations are mostly positive (corresponding to stable and global attribution) and that one is a likable person (corresponding to internal attribution; Mehrabian 1994). In the case of a positive social outcome, this general view is affirmed and, consequently, people attribute the outcome to internal (vs. external), stable (vs. variable), and global (vs. specific) causes. Similarly, when the negative outcomes occur that people high in social avoidance motives fear, people are likely to attribute them to internal, stable, and global causes. Again, this pattern of causal attributions helps to maintain a consistent sense of self and the world, even if it is a negative one (Alden 1986). In line with these hypotheses, Schoch et al. (2015) found that social approach (but not avoidance) motives predict attributions of social acceptance, whereas social avoidance (but not approach) motives predict attributions of social rejection.

Figure 1 shows the hypothesized relationships between social motives, attributions and expectations and speed-dating participation. To summarize, as shown in Fig. 1, we hypothesize that social approach and avoidance motives predict speed-dating participation (Study 1). Regarding the mechanisms underlying the effects of social motives on speed-dating participation, we hypothesize that social approach and avoidance motives predict attributions, attributions predict expectations, and expectations, in turn, predict speed-dating participation (Study 2). Specifically, social approach motives predict attributions of a positive social outcome (i.e., acceptance in a speed-dating event), whereas social avoidance motives predict attributions of a negative social outcome (i.e., rejection in a speed-dating event). These different attributions should lead to different expectations for similar future situations (i.e., an upcoming speed-dating event). The more people attribute acceptance to internal, stable, and global causes, the more they should expect acceptance in a similar future situation. The more they attribute rejection to internal, stable, and global causes, the more they should expect rejection in a similar future situation. Finally, we expect to replicate the previous finding that expectations for an upcoming social situation predict people’s willingness to enter this situation (Anthony et al. 2007).

Fig. 1
figure 1

The relationships among variables to be examined in study 1 and 2

Note, that we do not expect that social approach motives predict attributions of rejection and that social avoidance motives predict attributions of acceptance. This assumption is based on the existing empirical evidence demonstrating that social approach motives are particularly predictive in positive social situations, whereas social avoidance motives are particularly predictive in negative social situations (Gable and Poore 2008; Nikitin et al. 2012; Schoch et al. 2015; Strachman and Gable 2006).

Study 1

Method

Participants

Participants were recruited through an e-mail service for students. The sample consisted of college students between 18 and 30 years. N = 258 participants completed the online questionnaire. Data of 53 participants were excluded because these participants were not single. The final sample consisted of N = 205 students (age M = 23.18 years, SD = 3.23, 49% women). Very few of the participants had previous experience with speed dating (1.5%).Footnote 2

Procedure

After providing written informed consent, a sample of college students who were currently single completed an online questionnaire assessing their social approach and avoidance motives (run by http://www.limesurvey.org). After completing the questionnaire, the students decided whether––as a way of compensating participation in the study––they wanted to take part in a speed-dating event (that was conducted a few weeks later by our lab). To prevent a possible selection bias, speed dating was not mentioned before completion of the questionnaire. In addition, all participants could take part in a raffle for three train vouchers worth 50 Swiss francs each (approximately 51 USD). After participation, they were fully debriefed. Based on the checklist to self-assess studies concerning their ethical safety from the ethics committee of the local university, the present research was classified as safe and exempt from further review.

Measurement of social approach and avoidance motives

We used the German version of the Mehrabian affiliative tendency and sensitivity to rejection scales to assess social approach and avoidance motives (Mehrabian 1970; German version: Sokolowski 1986). A sample item of the Affiliative Tendency Scale is “I like to make as many friends as I can.” A sample item of the rejection sensitivity scale is “I prefer not to go to a place if I know that some of the people who will be there don’t like me.” The scales are a commonly used tool for assessing dispositional social approach and avoidance motives, and has good reliability and validity (e.g., Gable et al. 2003; Nikitin et al. 2012; Strachman and Gable 2006). For reasons of economy, we used a short version of the scales, each with eight items of the highest loadings based on a previous study (Nikitin and Freund 2010, Study 1; see Table S1 in the Supplemental Materials for the factor analysis). Response scales ranged from 0 (strongly disagree) to 6 (strongly agree). The internal consistency was Cronbach’s α = 0.72 for approach motives (M = 3.70, SD = 0.93) and α = 0.61 (M = 3.19, SD = 0.86) for avoidance motives. Approach and avoidance motives were not significantly correlated (r = − 0.10, p = 0.18).

Measurement of decision for/against participation in speed dating

Participants had the possibility to participate in a speed-dating event after completing the questionnaire. The instruction was as follows: “The questionnaire is now completed. Thank you very much for your participation! As a thank-you, you have the possibility to take part in a responsibly organized, free-of-charge speed-dating event worth 70 Swiss francs with other college students aged between 18 and 30 years.” For those who were not familiar with speed dating, a short description of the procedure was added. Participants were requested to click “Yes” or “No” dependent on whether they wanted to participate or not. About one-third (n = 67) of the participants decided to participate in the event.

Results

We ran a logistic regression analysis predicting the decision for or against participating in the speed-dating event by approach and avoidance motives. As hypothesized, avoidance motives negatively predicted the decision (B = − 0.55, SE = 0.19, Wald = 8.37, p = 0.004, OR 0.58, 95% CI [0.40, 0.84]), whereas approach motives marginally positively predicted the decision for participating in the speed-dating event (B = 0.34, SE = 0.18, Wald = 3.66, p = 0.06, OR 1.41, 95% CI [0.99, 1.9]).

Study 2

Study 1 supported our hypothesis that social approach and social avoidance motives differentially predict the decision to participate in a speed-dating event. Study 2 was conducted to test the predicted underlying mechanisms of these associations: attributions of previous experience with speed dating and expectations for future speed-dating events. Similar to Study 1, participants completed an online questionnaire and decided at the end whether––as a compensation for participating in the study––they wanted to take part in an actual speed-dating event. In addition, Study 2 manipulated acceptance and rejection in a hypothetical speed-dating situation and assessed participants’ attributions of this outcome. We decided to use a hypothetical scenario in order to comply with the APA guidelines to use deception only as a last resort if no other procedures are available. In addition, a manipulation using hypothetical scenarios was found to be viable in previous studies (e.g., Schoch et al. 2015). Based on the checklist to self-assess studies concerning their ethical safety from the ethics committee of the local university, the present research was classified as safe and exempt from further review.

Method

Participants

Participants were recruited through the participant server at the university. We approached students between 18 and 30 years who were currently single. N = 182 participants completed the online questionnaire. Data of 26 participants were excluded because these participants did not fulfill the above criteria. Another three participants were excluded because they failed to respond correctly to one of two control questions (detailed information is provided below). The final sample consisted of N = 153 single young adults (age M = 22.42 years, SD = 2.70, 73% women).Footnote 3 Very few of the participants had experience with speed dating (3.9%) (See footnote 2).

Procedure

Participants first gave written informed consent and then filled out the questionnaire on social motives. Subsequently, participants read a hypothetical speed-dating scenario that included a manipulation of acceptance or rejection, respectively. After the manipulation, participants rated the causes of the acceptance or rejection in the hypothetical speed-dating scenario on the attributional dimensions of internality, stability, and globality. Finally, participants rated specific positive and negative expectations concerning an actual upcoming speed-dating event and they decided if they wanted to take part in it. The possibility to participate in the actual speed-dating event was not mentioned before. An official speed-dating provider associated with the university conducted the actual speed dating a few weeks later. As an additional compensation, participants could take part in a raffle for 10 book vouchers worth 10 Swiss francs each (approximately 10 USD). Descriptive statistics of the main variables of Study 2 and mean differences between the acceptance and the rejection conditions are presented in Table 1. Bivariate correlations among all main variables are presented in Table 2.

Table 1 Descriptive statistics of main variables in study 2 with mean differences between the acceptance and the rejection condition
Table 2 Bivariate correlations among the constructs of study 2

Measurement of social approach and avoidance motives

As in Study 1, the Mehrabian affiliative tendency and sensitivity to rejection scales assessed social approach and avoidance motives. As the reliability of the avoidance motive subscale was not optimal in Study 1, we used all items of the scales (25 each) in Study 2. Cronbach’s α was 0.79 for the Affiliative Tendency Scale and was 0.82 for the Sensitivity to Rejection Scale (Cronbach’s α for each experimental condition can be found in Table 2).

Manipulation of social acceptance and rejection

Acceptance and rejection were manipulated via a hypothetical speed-dating situation. Schoch et al. (2015) demonstrated that hypothetical scenarios can be used to effectively manipulate participants’ perceived acceptance and rejection in social situations. Participants read the description of a speed-dating procedure and were then asked to imagine taking part in a speed-dating event: “Please imagine having participated in a speed-dating event. Over the course of this event, you got to know 10 persons of the opposite sex. Each date lasted 7 min. You liked four of these persons. You would like to meet them again and selected them for further contact. You know that you only get their contact information if there is a match between your and your partner’s choice. Now you are waiting eagerly for the result pondering the questions ‘How many persons might have chosen me?’ and ‘How many matches do I have?’” After reading this instruction, participants were randomly assigned to receiving either a fictitious acceptance feedback (“You wanted to get to know four persons––all of them also wanted to get to know you.”; n = 78) or a fictitious rejection feedback (“You wanted to get to know four persons––none of them wanted to get to know you.”; n = 75). To ensure that participants engaged in imagining the speed-dating situation, we asked them to take some notes about their thoughts and hypothetical behavior after each step of the manipulation.

Manipulation check

To test whether the manipulation of social acceptance and rejection was successful, we asked participants how successful the imagined speed dating was for them. Participants responded on a scale ranging from 0 (not successful at all) to 6 (very successful). Participants reported significantly more success in the acceptance condition (M = 5.60, SD = 0.67) than in the rejection condition (M = 0.94, SD = 1.21), t(151) = 29.63, p < 0.001, d = 4.82.

Measurement of attributions

An adapted version of the Attributional Style Questionnaire for Adults (Poppe et al. 2005) assessed attributions of acceptance and rejection, respectively (see also Schoch et al. 2015). After the hypothetical feedback, participants wrote down the main cause of their success or failure in the speed-dating scenario. They rated this cause on the dimensions of internality (e.g., “The cause has something to do with circumstances or other persons” vs. “The cause has something to do with me”), stability (e.g., “The cause will change over time” vs. “The cause will be stable over time”), and globality (e.g., “The cause applies only to this situation” vs. “The cause applies to many other situations”). Each dimension was assessed with two single items. Response scales ranged from − 3 (external, variable, specific) to + 3 (internal, stable, global). Cronbach’s α was 0.78, 0.51, 0.77, and 0.83 for the internality dimension, stability dimension, globality dimension, and the whole scale, respectively (Cronbach’s α for each experimental condition can be found in Table 1).

Measurement of specific expectations

Specific expectations of an upcoming actual speed-dating event were assessed with eight self-created items. Four items assessed positive expectations (“I expect …” “… that others will accept me”, “… that others will like me”, “… to establish contact with many people”, “… to get on well with others”), four items assessed negative expectations (“I expect …” “… that others are not interested in me”, “… that others will reject me”, “… that hardly anybody will choose me”, “… to be rejected”). Cronbach’s α was 0.81 for the positive expectations and 0.87 for the negative expectations (Cronbach’s α for each experimental condition can be found in Table 2).

Measurement of decision for/against participation in speed dating

The instruction regarding the decision for or against participating in an actual speed-dating event was the same as in Study 1. In the acceptance condition, 26 of 78 (33%) participants decided for attending the speed dating event, and 30 of 75 (40%) in the rejection condition. The difference between acceptance and rejection condition was not statistically significant, χ2(1153) = 0.39, p = 0.41.

Control items

One control item (“How many persons did you select for further contact?” asked directly after the manipulation of acceptance/rejection) identified participants who did not read the hypothetical scenario carefully. Participants who did not respond correctly (n = 3) were excluded from the analyses.

To identify how realistic the speed-dating scenario was, we used two items (“How realistic was the speed-dating scenario for you?”, “How well could you imagine the speed-dating scenario?”). Response scales ranged from 0 (not at all) to 6 (very). Participants experienced the hypothetical speed-dating scenario as more realistic than unrealistic (M = 3.56, SD = 1.43) and could imagine it relatively well (M = 4.38, SD = 1.30). Sample means were significantly above the scale center of 3 (realism: t[152] = 4.86, p < 0.001, d = 0.79; imaginability: t[152] = 13.09, p < 0.001, d = 2.12). We regressed the perceived realism and imaginability respectively on experimental condition, approach social motives, and negative social motives. The results showed that none of the main effects of condition, main effects of social motives, or their interaction effects was significant (ps > 0.09), which indicate that realism and imaginability did not vary with condition or social motives.

Results

Data-analytical strategy

We included social approach and avoidance motives, attributions, positive expectations, and negative expectations in a multiple-group latent structural equation model for social acceptance (Group 1) and social rejection (Group 2) to test whether the hypothesized motives–attributions–expectations link differs between the two conditions (see Fig. 2). In order to reduce the number of parameters, we used the item parceling method (Little et al. 2002) to build the latent variables for social approach and avoidance motives (e.g., Elliot and Thrash 2002; Marsh et al. 1998; Neria et al. 2016; Wang et al. 2017). Similarly, the three attribution dimensions internality, stability, and globality were aggregated into one latent variable, a widely used procedure in attributional research (e.g., Ciarrochi and Heaven 2008; Hawkins and Miller 2003; Sweeney et al. 1986). Finally, the four manifest variables of the positive-expectations scale were used to build the latent variable “positive expectations,” and the four manifest variables of the negative-expectation scale were used to build the latent variable “negative expectations” (see Table 3 for the loadings of the manifest variables on the latent variables). All variables were z-standardized, in order to reduce the effects of cross-variable variation in measure unit and multicollinearity among these variables, before they were included in the model. Analyses were conducted using lavaan (Rosseel 2012), an R-based (R Development Core Team 2012) package for structural equation modeling.

Fig. 2
figure 2

Structural equation model with standardized estimated values of the relationship between social motives, the composite attribution score (internal vs. external, stable vs. variable, and global vs. specific attributions), and specific expectations after the acceptance scenario (upper line) and the rejection scenario (lower line). All coefficients (except for those on the dashed paths) are standardized, and their corresponding standard errors are presented in parentheses. The dashed paths are not part of the structural equation model (because speed-dating participation is a nominal measure, which, if treated as an endogenous variable, would cause problems for model estimation) but for presentation purpose only. The coefficients and standard errors (in parentheses) on the dashed paths are estimated by logistic regression, and they are not significantly moderated by condition (i.e., acceptance vs. rejection scenario). *p < 0.05, **p < 0.01, ***p < 0.001

Table 3 Standardized estimates of the loadings of the manifest variables on the latent variables in study 2

Speed-dating participation decision was not included in the model because it was a nominal measure (yes or no). To our knowledge, there is no well-established method or statistical package to handle complex mediation or structural equation models with nominal outcome (or endogenous) variables. Thus, we tested for our hypotheses with two separate models: First, we built a structural equation model to examine the relationship among social motives (social approach and avoidance motives), attributions, and expectations (positive expectations and negative expectations); Second, we employed a binary logistic regression model to test the relationship between positive/negative expectations with decision for/against speed-dating participation.

Social motives and attributions as predictors of specific expectations

The multiple-group latent structural equation model yielded fair goodness-of-fit indices, χ2 (df = 216, n1 = 78, n2 = 75) = 247.30, p = 0.07, CFI = 0.94, TLI = 0.93, RMSEA = 0.044, SRMR = .071, indicating that the model overall well fit to the data of the acceptance and rejection conditions (e.g., Schreiber et al. 2006). Satisfactory goodness-of-fit indices were obtained if we modelled the data separately for the acceptance (χ2 = 159.24, df = 109, p = 0.001, CFI = 0.96, TLI = 0.95, RMSEA = 0.054, SRMR = 0.075) and rejection condition (χ2 = 170.77, df = 109, p < 0.001, CFI = 0.94, TLI = 0.93, RMSEA = 0.052, SRMR = 0.075). We also built an alternative multiple-group latent structural equation model, in which path coefficients were constrained to be equal between Group1 and Group 2. The alternative model showed relatively poorer goodness of fit indices, χ2 (df = 224) = 285.62, p = 0.003, CFI = 0.89, TLI = 0.87, RMSEA = 0.060, SRMR = 0.089, compared to the former multiple-group latent structural equation model (Δχ2 = 38.32, Δdf = 8, p < 0.01). The results indicated that the relationships among social motives, attributions, and expectations significantly differed between the acceptance condition (Group 1) and the rejection condition (Group 2).

In the acceptance condition, as hypothesized, social approach motives predicted attributions of acceptance, which, in turn, positively predicted specific positive expectations and negatively predicted negative expectations for the upcoming speed-dating event (see Fig. 1, upper line). In other words, the higher the approach motives were, the more participants tended to attribute the imagined acceptance in the speed-dating scenario to the composite score of internal (vs. external), stable (vs. variable), and global (vs. specific) causes. In turn, the more participants attributed acceptance to internal, stable, and global causes, the more they had positive expectations and the less they had negative expectations for an upcoming speed-dating event. Social avoidance motives did not significantly predict attributions of acceptance. Instead, social avoidance motives directly negatively predicted positive expectations and positively predicted negative expectations. The higher avoidance motives were, the lower were positive expectations and the higher were negative expectations for an upcoming speed-dating event.

In the rejection condition, social avoidance motives predicted attributions of rejection, which, in turn, predicted expectations for the upcoming speed-dating event (see Fig. 1, lower line). In other words, the higher the avoidance motives were, the more participants tended to attribute the imagined rejection in the speed-dating scenario to the composite score of internal (vs. external), stable (vs. variable), and global (vs. specific) causes. In turn, the more participants attributed rejection to internal, stable, and global causes, the less positive expectations and the more negative expectations they had for an upcoming speed-dating event. Social approach motives did not significantly predict attributions of rejection. Instead, social approach motives directly predicted positive expectations but not negative expectations. The higher approach motives were, the higher were positive expectations for an upcoming speed-dating event.

Social motives and expectations as predictors of speed-dating participation

As discussed above, it is inappropriate to incorporate speed-dating participation, a binary nominal dependent variable, into the structural equation models. We thus performed binary logistic regression analyses to examine the relationships between speed-dating participation and its predictors. First, we regressed speed-dating participation on social approach and avoidance motives. The results showed that the effects of social approach and avoidance motives, as well as the interaction between social motives and condition, were non-significant (ps > 0.21). Second, we regressed speed-dating participation on negative and positive expectations. The results showed that speed-dating participation was significantly predicted by positive expectations, B = 0.93, SE = 0.37, Wald = 6.23, p = 0.01, OR = 2.54, 95% CI [1.22, 5.30] but not by negative expectations, B = 0.48, SE = 0.29, Wald = 2.80, p = 0.09, OR 1.62, 95% CI [0.92, 2.86]. These effects were not moderated by experimental condition (ps > 0.38). Overall, these results demonstrate a significant indirect relationship between social motives and speed-dating participation via attributions and positive expectations, despite the non-significant direct relationship.

Supplementary analyses

The results of the structural equation modeling and binary logistic regression, altogether, suggest significant indirect effects of social approach and avoidance motives on speed-dating participation via attribution and positive expectations despite non-significant total effects (i.e., the sum of indirect and direct effects). In order to more clearly illustrate the indirect effect of social motives on speed-dating participation via attributions and expectations, we built a comprehensive structural equation model (with speed-dating participation included as the outcome variable) as a supplementary analysis. In this model, we treated speed-dating participation as an ordered dichotomous measure (0 = no, 1 = yes; the higher score indicates a higher level of likelihood to participate in the speed dating), such that the R-based package lavaan would be able to estimate the model (Rosseel 2012). The results showed that speed-dating participation was significantly predicted by positive expectations, positive expectations were significantly predicted by attributions, and attributions were significantly predicted by social approach or avoidance motive (see Fig. S1 in the Supplemental Materials). These results additionally support the indirect effect of social motives on speed-dating participation.

General discussion

Although there exists a large body of research on the initiation of romantic relationships, the present two studies are the first to test the role of dispositional motives. These studies demonstrated that social approach and avoidance motives and their social-cognitive concomitants (i.e., attributions, expectations) play an important role for the decision to participate in a speed-dating event, a sample case of relationship initiation.

Overall, the current studies contribute to our understanding of motivational underpinnings of relationship initiation for research areas that are directly or indirectly associated with motivation. For example, the present findings show the need to specify the cognitive processes associated with high versus low self-esteem. Assuming that high self-esteem is associated with high social approach motivation and low self-esteem with high social avoidance motivation, cognitive processes that lead to the initiation of a relationship in people high in self-esteem seem to differ from those low in self-esteem (see also Stinson et al. 2014). Similarly, the present findings contribute to research on attachment styles that seem to be differentially associated with social approach and avoidance motives (e.g., Nikitin and Freund 2010): whereas secure attachment is associated with a combination of high approach and low avoidance motives, insecure-fearful attachment is associated with a combination of high approach and high avoidance motives, and insecure-avoidant attachment with low approach motives. Based on the findings of the current studies, the three attachment styles might lead to relationship initiation through different cognitive mechanisms (for supporting results on avoidant attachment see Spielmann et al. 2013).

Specifically, the current studies provide three main insights on the role of motives for relationship initiation: First, motive-related attributions are differential. In line with previous research showing that social approach motives are particularly predictive in positive social situations and that social avoidance motives are particularly predictive in negative social situations (Gable and Poore 2008; Nikitin et al. 2012; Strachman and Gable 2006), social approach motives in the present study predicted attributions of acceptance, whereas social avoidance motives predicted attributions of rejection (Schoch et al. 2015). These results support the notion that social approach and avoidance motives are two independent motivational systems that influence social behavior and experience through different processes (Gable and Berkman 2008; Nikitin and Freund 2008). Thus, to predict and explain behavior and experience in social relationships, both motives should be considered. As Gable and Poore (2008) put it, researchers may sometimes be comparing apples and oranges when examining the same outcome across people. For some people, deciding to enter a situation that offers the possibility to initiate a romantic relationship might originate mainly from the experience of previous positive experiences, for others mainly from the experience of previous negative experiences. This has an important implication for intervention strategies aiming at changing people’s decisions to enter social situations. Such attempts might be only successful when considering the differential paths leading to this decision.

Social approach motives were directly associated with (positive but not negative) expectations for a future speed-dating event in the rejection condition, and indirectly associated with (positive and negative) expectations via attributions in the acceptance condition. In contrast, social avoidance motives were directly associated with (positive and negative) expectations for a future speed-dating event in the acceptance condition, and indirectly associated with (positive and negative) expectations via attributions in the rejection condition. These differential associations suggest that social motives modulate attributions, and attributions, in turn, impact expectations in situations that are in focus of the motives (i.e., an acceptance situation for social approach motives and a rejection situation for social avoidance motives). Although not predicted, these associations also illustrate how social motives operate in situations that are not in focus of the motives. It seems that in such situations, the available information (acceptance or rejection) is not integrated in the resulting evaluation (i.e., attributions) of the situation. Thus, the adaptiveness of high social approach motives might be based both on crediting oneself for social acceptance and “neglecting” social rejection. The low adaptiveness of high social avoidance motives might be based both on blaming oneself for social rejection and “neglecting” social acceptance. These post-hoc interpretations require to be directly tested in future studies.

Third, the present research illustrates possible antecedents of different attributional styles. Despite the long tradition of attributional research, relatively little attention has been devoted to the individual precursors of attributions (with exception of psychopathology; see a meta-analysis by Mezulis et al. 2004). Based on the present findings, it seems that the dispositional motivation to approach positive social outcomes and the dispositional motivation to avoid negative social outcomes are important antecedents of individual differences in attributions of social events. Although people generally tend to attribute positive events to internal, stable, and global causes, and negative events to external, variable, and specific causes (Mezulis et al. 2004), this attributional “bias” seems to be modulated by people’s motives. We argued previously that the association between social approach and avoidance motives and attributions is driven by the general view or, in other words, generalized (positive or negative) expectations that people have in social situations (Mehrabian 1994). A person who generally expects that he or she succeeds in social situations will feel affirmed by a positive social outcome. A person who generally expects that he or she fails in social situations will feel affirmed by a negative social outcome. There is substantial research on the effects of attributions on expectations (e.g., Anderson and Jennings 1980; Brodt and Zimbardo 1981; Sedikides and Alicke 2012). In contrast, little is known about the role of expectations for attributions (Alden 1986). The present findings support the notion that general expectations as operationalized by social approach and avoidance motives play an important role for individual differences in attributions of social events (Schoch et al. 2015).

There were several unexpected findings in the present studies. First, there was indirect association, but no direct association between social motives and the decision to participate in the speed-dating event in Study 2. There is some evidence that the power of the test of indirect effects is often considerably greater than the power of the test of the direct effect, even when they are both of the same magnitude (Kenny and Judd 2013). This might explain why the indirect, but not the direct effect was statistically significant. In addition, Study 1 that found the predicted direct effect of motives on the decision to participate assessed the decision directly after assessing the motives; this was not the case in Study 2, which manipulated participants’ perceived acceptance and rejection before assessing the decision and, thus, might interrupt the direct effects of social motives on speed-dating participation.

Second, negative expectations did not predict the decision to participate in a speed-dating event. In other words, participants decided for or against participation in a speed-dating event irrespective of how strongly they expected being rejected in the speed-dating event. This is surprising given the role of negative expectations for behavior (Leder et al. 2013). The lack of an association between negative expectations and the decision to participate in a speed-dating event might also be explained by the specific context of the speed-dating event. Rejection in the speed-dating event is not public. When there is no match between interests, only the person that was interested in the other person would know that the other person did not choose him or her. Although this disinterest might still hurt, the rejected person saves his or her face (Beck and Clark 2009). In addition, rejection in the speed-dating is not face-to-face but takes place indirectly by not electing somebody for a date. This likely makes the rejected person feel less awkward and stressed (compared to face-to-face rejection). Moreover, rejection in the speed-dating is typically from an unfamiliar person, which might not be perceived as a loss (e.g., “I do not know her anyway, so I will lose nothing if she rejects me”). Another possible explanation is that expecting acceptance is generally more powerful for the decision to enter a social situation than expecting rejection (Anthony et al. 2007; Cameron et al. 2010; Gable 2006; Nikitin et al. 2012). In other words, not expecting rejection is not a strong incentive to enter a social situation. Instead, one has to expect rewards in the situation to enter it.

Limitations

The current studies have several methodological limitations. First, in Study 2 the sex ratio was skewed: there were more female than male participants. However, our analyses (see Footnote 3) showed that sex did not impact the central variables (i.e., approach social motives, avoidance social motives, attributions, positive and negative expectations, and speed-dating participation) or the relationships among these variables.

Second, the decision for/against speed-dating participation was measured as a binary nominal variable, which limited the use of sophisticated statistical models, such as the structural equation model. Future study could consider measure this variable on an ordinal or continuous scale. One possibility is to ask participants to rate the strength of their willingness to partake a speed-dating on a multipoint response scale.

Third, we manipulated the outcome of a hypothetical speed-dating situation to investigate the effects of acceptance and rejection on subsequent actual speed-dating participation. This manipulation was found to be effective in previous research (Schoch et al. 2015), and it also significantly moderated the motivational and cognitive processes underlying initiation of speed dating in the current study. However, the effects of an imagined scenario may be less strong than the effects of an actual event, which might result in the absence of difference in the decision for a speed-dating participation between the acceptance and rejection conditions in the current study. Future studies could consider use stronger manipulations of social acceptance and rejection.

Another limitation is that it is unclear whether the present findings can be generalized to other situations of relationship initiation than speed dating. Eastwick and Finkel (2008) demonstrated that speed-dating procedures incorporate all features of an optimal paradigm to test relationship initiation such as study of real relationships with a potential future, the possibility to implement experimental manipulations, the possibility to give participants multiple romantic options or to get background characteristics before participants meet. Despite these strengths, speed-dating events are different from other relationship initiations. For example, as discussed above, the imagined rejection in a speed-dating event differs from many other social situations as nobody but the involved person him- or herself knows the outcome. A replication of the present findings in different contexts of relationship initiation seems important in order to test their generalizability.

Conclusions

The present studies show that the integration of motivational and attributional research might be fruitful to explain decisions for relationship initiation. The main insights for the study on social relationships are that (1) what people generally desire and fear in social relationships are two important antecedents of the decision to initialize a romantic relationship; (2) these generalized desires and fears have different social-cognitive mechanisms (i.e., attributions, expectations); in other words, different social-cognitive mechanisms might underlie the same overt behavior (i.e., the decision to participate in a speed-dating).