When your favorites disappoint: Self-construal influences response to disappointing brand experiences

Abstract

This research examined the effect of self-construal (independent vs. interdependent) on people’s responses to disappointing brand experiences. We propose that following a disappointing brand experience, independents are more likely than interdependents to express intentions for negative behavior. This effect is due to differences in the importance ascribed to various goals: independents are focused on their own personal goals and expectations, whereas interdependents are focused on maintaining pleasant and harmonious relationships. Consequently, when independents experience disappointment (i.e., their expectations are not met), they appraise the situation as less pleasant than do interdependents. Independents are thus more likely to experience negative emotions, which in turn lead to negative behavioral intentions. Three studies, in which self-construal was primed, supported this prediction. In experiment 1 participants imagined a sports event where their favorite team played carelessly and lost. Participants in the independent (vs. interdependent) prime condition were more likely to express intentions of negative behavior toward the sports team; negative emotions partially mediated this effect. Experiment 2 provides evidence for the mediating role of emotional appraisal in the extent to which the disappointing experience (sports event) is perceived as unpleasant. Experiment 3 replicates these findings in the context of service failure at a restaurant. It also provides evidence for the role of prior expectations in this effect, demonstrating that the effect occurs only when participants have prior expectations, and does not occur in their absence. This research sheds light on the effects of self-construal on emotional and behavioral responses to negative brand experiences, and highlights the processes underlying these effects.

Introduction

Despite firms’ best efforts, consumers often leave brand interactions feeling a “I didn’t expect that from X” reaction that characterizes disappointing brand experiences (Tilley, 1999). Brand disappointment is defined as negative feelings resulting from the perception that a brands’ performance has failed to meet consumers’ expectations (Zeelenberg & Pieters, 1999). Consumers’ brand expectations are shaped by strategically designed marketing messages, and also by ongoing brand interactions, which lead to the formation of consumer-brand attachments across a wide variety of branded goods, services, and experiences, including food, apparel, travel destinations, and sports teams, just to name a few (see, e.g., Belk, 1988; Fournier, 1998; Kaynak et al., 2008).

Importantly, consumers’ disappointing experiences can impact future behavior toward the brand (Zarantonello et al., 2016). For example, whereas after a winning football game, American students tend to exhibit behavior that expresses their affiliation with their favorite team (Cialdini et al., 1976), the opposite occurs when the team loses (Bizman & Yinon, 2002). Other negative brand experiences evoke similar reactions. For example, disappointing service encounters lead consumers to switch providers (Keaveney, 1995) and to negative word-of-mouth (Philp et al., 2018). Moreover, consumers may experience brand failures as personal failures, which could harm their self-view (Cheng et al., 2012). Such negative effects of brand failures may take place not only upon involvement in a disappointing experience, but also when the consumer has been merely exposed to negative brand information (Aaker et al., 2004; Swaminathan et al., 2007).

In this research we examined factors involved in the effects of brand disappointment. Specifically, we argue that negative reactions to disappointing brand experiences are contingent upon people’s salient self-construal view: whether they view themselves as separate from others (i.e., independent) or as part of a group (i.e., interdependent; Markus & Kitayama, 1991; Triandis, 1989). Because people with independent (vs. interdependent) self-construal tend to focus on fulfilling their own goals and expectations (Van Horen et al., 2008), upon encountering a disappointing experience (i.e., unmet expectations) they focus on the self-relevant implications of the experience (i.e., they will take them more personally). Independents, therefore, are likely to appraise the disappointment as unpleasant, to experience negative emotions (Bonifield & Cole, 2007), and consequently, to express negative behavioral intentions toward the brand. In contrast, interdependents, who are more focused on maintaining harmonious relationships with others and less focused on their own goals and expectations (Heine et al., 1999), are less likely to appraise disappointing brand experiences as unpleasant, to experience negative emotions, and to express negative behavioral intentions toward the brand.

We conducted three studies to examine our conceptualization. Experiment 1 demonstrated that participants under an independent prime condition were more likely than those under an interdependent prime condition to experience negative emotions when their favorite basketball team played carelessly and lost; the negative emotions, in turn, led those in the independent condition to express negative behavioral intentions. Experiment 2 showed that participants in the independent prime condition appraised the disappointing event as more unpleasant than those in the interdependent condition, leading to negative behavioral intentions. Experiment 3 replicated these findings in a service context, and supported the premise that the distinct responses of independents (vs. interdependents) are due to their focus on their unmet expectations.

Self-Construal and Responses to Disappointing Brand Experiences

Researchers interested in the effects of negative brand experiences have examined negative responses such as a desire for vengeance (Bechwati & Morrin, 2003), negative word of mouth (Philp et al., 2018) and complaining (Bearden & Teel, 1983). Yet, Bonifield and Cole (2007) pointed out that much of the extant literature has overlooked the role of emotions. To address this deficiency, Bonifield and Cole highlighted the role of anger in explaining negative versus positive post-purchase behaviors. Our research contributes to this knowledge by highlighting the moderating role of self-construal on consumers’ emotional and behavioral responses to negative experiences.

Self-Construal

A large body of research has dealt with the complexity of the self, which manifests itself in multiple possible representations (e.g., Baumeister, 1986; Greenwald & Pratkanis, 1984; James, 1890). In accordance with this line of thought, Triandis (1989) as well as Markus and Kitayama (1991) argued that people in certain cultures share salient self-views. Triandis (1989) distinguished between the private self, public self, and collective self and suggested that people in different cultures vary in the expression of the various selves through sampling. Markus and Kitayama (1991) proposed that cultural background leads people to construe the selves differently, and coined the term self-construal to refer to self-view of people in individualistic and collectivist cultures. The researchers distinguished between independent and interdependent self-construals, which are salient in individualistic and collectivist cultures, respectively (Markus & Kitayama, 1991). People with salient independent self-construal view themselves as autonomous, separated from other people, and distinct from the ingroup. They ascribe high importance to uniqueness and to their own goals, expectations, achievements, and accomplishments. By contrast, people with salient interdependent self-construal view themselves as linked to other people. They ascribe high importance to conformity, group harmony, relationships with others, and to the goals of the group.

Although cultural contexts promote the development of specific self-construal, leading to cultural differences in chronic self-construal, people in all cultures generally possess both independent and interdependent tendencies, each of which can be activated in various contexts (Triandis, 1995). Yet, because independent and interdependent self-construals are more salient in individualist (Western) and collectivist (non-Western) cultural contexts, respectively, researchers have used various self-construal operationalizations to examine the differences between individualist and collectivist cultures. Various approaches have been used to examine the effects of self-construal across the independent-interdependent dimension (Cross et al., 2011), comparing Western (e.g., Americans) and non-Western (e.g., Asians; Ji et al., 2000) cultures known to be individualist-independent and collectivist-interdependent respectively; measuring self-construal within a specific culture (Riemer & Shavitt, 2011); or manipulating cultural self-construal (Lalwani & Shavitt, 2009; Zhang & Shrum, 2009). Numerous studies examined the effect of self-construal on people’s cognition, motivation, and emotion, and indeed many of the studies demonstrate consistent findings when self-construal was operationalized using the various approaches (for a review see Cross et al., 2011).

Self-Construal and Responses to Brand Disappointments

Swaminathan et al. (2007) have examined effects of individual (independent) versus group-level (interdependent) self-construal on brand evaluations following exposure to negative brand information. A key contribution of Swaminathan and colleagues is highlighting the path through which consumers-brand connections influence their responses to negative brand information. Our goal in the current research was to enhance the understanding of the effect of self-construal on consumers’ responses to negative brand experiences by uncovering the emotional processes underlying these effects.

We argue that three interrelated factors and processes associated with independent and interdependent self-construals may be involved in consumers’ responses to disappointing experiences. First, consumers with independent versus interdependent self-construal may differ in the extent to which they perceive negative events as self-relevant. An independent self-construal is associated with people viewing themselves as autonomous and separated from the group. For independents, personal goals and expectations are superior to those of the group, and social behavior is guided by personal desire (Escalas & Bettman, 2005; Heine et al., 1999; Riemer et al., 2014). Consequently, we suggest, when a negative event occurs, independents’ attention will be drawn to the personal implications of the negative event. By contrast, interdependent self-construal is associated with people viewing themselves as connected to others (Mao et al., 2016). Interdependents are committed to the group; for them, personal goals and expectations are inferior to those of the group (Escalas & Bettman, 2005; Riemer et al., 2014), and they tend to focus on maintaining social harmony (Heine et al., 1999). Thus, when a negative event occurs, interdependents are more likely to focus on its implications to their ingroup.

Second, appraisal theories of emotions suggest that emotions are elicited and shaped by people’s subjective evaluations of an antecedent situation (Lazarus, 1991; Roseman, 1991; Scherer, 1999; Smith & Ellsworth, 1985). The subjective nature of the evaluation explains why similar experiences can stimulate different emotions in different people. Due to the subjective nature of situation evaluation, and to differences in values and thinking styles between people of various self-construals (e.g., Nisbett et al., 2001), we expect differences in the appraisal of disappointing experiences. Specifically, we predict that disappointing events will be appraised as less pleasant by people with independent (vs. interdependent) self-construal, as their focus is on the self. Such differences in emotional appraisals should lead to differences in the interpretation of the situation, resulting in distinct responses.

Third, people with interdependent self-construal possess a holistic thinking style, which, among other things, enhances dialectical thinking (i.e., the ability to accept contradictions; Peng & Nisbett, 1999; Nisbett et al., 2001). Consequently, interdependents can see positive aspects in negative events (Hoffman, 1994; Williams & Aaker, 2002), and are therefore less likely to view disappointing experiences as unpleasant, to experience negative emotions after such events, and, thus, to express intentions for negative behaviors.

Altogether, these factors and processes should lead to differences in the extent to which independents and interdependents will respond negatively to disappointing brand experience. Specifically, we predict:

• H1: Following a disappointing brand experience, independents will be more likely than interdependents to experience negative emotions, which in turn will lead to negative behavioral intentions towards the brand.

Our conceptualization suggests that independents’ negative reactions to brand disappointments are driven, in large part, by their differential focus on their own unmet expectations. If this is the case, then the existence of prior expectations is expected to strengthen the effect, whereas in the absence of prior expectations, the effect will not occur, as follows:

• H2a: When prior expectations exist, independents (vs. interdependent) are more likely to react negatively to disappointing experiences.

• H2b: In the absence of expectations, the difference between independents and interdependents in their reactions to disappointing experiences is diminished.

Overview of Studies

To examine causality in the relationship between self-construal and response to disappointing brand experiences, we used manipulations of self-construal (Oyserman & Lee, 2008) in our studies. Self-construal priming procedures rely on the premise that despite the chronic accessibility of certain self-construal in a given cultural context, people in all cultures possess both independent and interdependent self-construals, and situational factors can make certain self-knowledge temporarily accessible (Cross et al., 2011).

The self-construal manipulation in our three studies consisted of a reading passage about an ancient Sumerian general, who had to choose a warrior. In the independent condition, the general chooses a talented warrior; in the interdependent self-construal condition, the general chooses a family member. After reading the passage, participants answered questions focusing on either the self or the other, for the independent and interdependent conditions respectively (Trafimow et al., 1991; Ybarra & Trafimow, 1998). Past studies have validated this manipulation either by showing that participants in the independent prime condition used more individual and private terms to describe themselves, while participants in the interdependent prime condition used more collective and group-oriented terms (e.g., Mandel, 2003; Ybarra & Trafimow, 1998), or by using the Singelis (1994) self-construal measure (e.g., Zampetakis et al., 2015).

Participants in our studies were randomly assigned to either independent or interdependent self-construal prime conditions, and were then exposed to scenarios of disappointing brand experiences. The first experiment examined the role of self-construal in people’s emotional and behavioral responses to a loss of a sports team. The second experiment investigated the mechanisms underlying the effect, exploring dialectical thinking and emotional appraisals. The third experiment replicated the first two experiments in the context of service failure at a restaurant, and examined the moderating role of prior brand expectations. All participants signed consent forms before each experiment as a prerequisite to participation. The protocols of the studies were approved by our university’s IRB.

Experiment 1: Responses to a Disappointing Sports Event

Method

Sixty undergraduate students (27% male, Mage = 23.5 years), who received extra course grade points for their participation, were randomly assigned to one of two self-construal conditions. After completing the self-construal manipulation, participants read a scenario describing a basketball game in which one’s favorite team played carelessly and lost (Appendix A), and then completed measures of their emotional and behavioral reactions.

To measure participants’ emotional reactions, the participants rated on a seven-point scale the extent to which the scenario made them feel each of 24 emotions (anger, eagerness, agitation, guilt, joy, surprise, anxiety, fear, interest, shame, quiescence, frustration, relief, contempt, pride, sadness, happiness, hope, disgust, contentment, dejection, elation, gratitude and cheerfulness; see Higgins, 1987; Richins, 1997; Smith & Ellsworth, 1985).

To measure behavioral intentions, the participants rated on a seven-point scale the extent to which they would tend to engage in each of ten behaviors. Five of the behaviors listed in this part were negative and five were positive (negative: “I will stop going to the team’s games”, “I will stop admiring the team due to its disrespect”, “I will condemn the way the coach runs the game”, “I will leave the court before the end of the game”,“I will ‘boo’ the players of the team”; positive: “I will continue to admire the team with devotion”, “I will give the players the respect they deserve despite their loss”, “I will continue to go to future games of the team”, “I will stay at the game until the last whistle”, “I will praise the coach at the end of the game”).

Results and Discussion

Participants in the independent prime condition expressed higher intentions for negative behavior than those in the interdependent condition (M_independent = 2.19, (SD = 1.06), M_{interdependent} = 1.60 (SD = 0.66), t(48.4) = −2.53, p = .014). To ensure that these results are not due to extreme responses bias, we also analyzed the positive behavioral intentions (in which we did not expect differences following negative experiences). Indeed, positive behavioral intentions did not differ across conditions (M_independent = 4.80 (SD = 1.15), M_{interdependent} = 5.18 (SD = 1.09), t(58) = 1.29, p = .20).

Participants in the independent (vs. interdependent) condition demonstrated a greater extent of negative emotions (M_independent = 4.02 (SD = 1.08), M_{interdependent} = 3.39 (SD = 1.13), t(58) = −2.19, p = .033). This was true for an aggregated measure of all negative emotions, and specifically for the following emotions: anger (M_independent = 4.80 (SD = 1.21), M_{interdependent} = 4.10 (SD = 1.65), t(53.34) = −1.87, p = .067), fear (M_independent = 3.63 (SD = 1.77), M_{interdependent} = 2.60 (SD = 1.79), t(58) = −2.25, p = .029), shame (M_independent = 4.53 (SD = 1.57), M_{interdependent} = 3.33 (SD = 1.77), t(58) = −2.780, p = .007), contempt (M_independent = 4.43 (SD = 1.72), M_{interdependent} = 3.57 (SD = 1.89), t(58) = −1.861, p = .068), and sadness (M_independent = 5.07 (SD = 1.51), M_{interdependent} = 4.10 (SD = 1.71), t(58) = −2.32, p = .024). The effects of self-construal on other negative emotions were insignificant: guilt (M_independent = 2.20 (SD = 1.24), M_{interdependent} = 1.93 (SD = 1.39), t(58) = −.78, p = .44), joy (M_independent = 2.10 (SD = .95), M_{interdependent} = 2.16 (SD = 1.14), t(58) = .24, p = .81),, anxiety (M_independent = 3.50 (SD = 2.04), M_{interdependent} = 2.90 (SD = 1.91), t(58) = −1.17, p = .25), frustration (M_independent = 5.13 (SD = 1.65), M_{interdependent} = 5.44 (SD = 1.70), t(58) = .46, p = .65), disgust (M_independent = 2.46 (SD = 1.73), M_{interdependent} = 1.83 (SD = 1.55), t(58) = −1.49, p = .14), and dejection (M_independent = 4.43 (SD = 1.65), M_{interdependent} = 4.26 (SD = 1.63), t(58) = −.39, p = .70). Positive emotional reactions did not differ between the conditions (M_independent = 2.72 (SD = 0.80), M_{interdependent} = 2.70 (SD = 0.93), t(58) = −.085, p = .93).

We also looked at the mediation role of negative emotions in the effect of self-construal on behavioral intentions. Self-construal (1 = interdependent, 2 = independent) was positively associated with intentions for negative behavior (c-path; B = .58, t(58) = 2.54, p < .05). Self-construal was positively related to negative emotions (a-path; B = .62, t(58) = 2.19, p < .05). Finally the mediator – negative emotions – was positively associated with intentions for negative behavior (b-path; B = .12, t(58) = 1.24, p < .05). Because both the a-path and b-path were significant, the model meets the criteria according to Baron and Kenny (1986). Mediation was also tested using a bootstrapping method with bias-corrected confidence estimates (Preacher et al., 2007). In the present study, the 95% confidence interval of the indirect effects was obtained with 5000 bootstrap resamples (Preacher & Hayes, 2008). Results confirmed the mediating role of negative emotions in the relation between self-construal and intentions for negative behavior (B = .08; CI = 0.01,0.34). The direct effect of self-construal on intentions for negative behavior remained significant (albeit less pronounced, c′-path; B = .49, t(58) = 2.11, p < .05) when controlling for negative emotions, thus suggesting partial mediation (Fig. 1).

Fig. 1

Indirect effect of self-construal on intentions for negative behavior through negative emotions (Experiment 1)

We ran additional conditions in which the scenario was positive. In this case we did not expect any differences, because both independents and interdependents would satisfy their main goal. Independents could maximize their positive emotions, and interdependents could have an enjoyable experience that includes social interaction (Heine et al., 1999). As expected, the findings showed insignificant differences between participants in the independent and interdependent conditions both in behavioral intentions and in emotional response (Negative behavior: M_independent = 1.15 (SD = 0.37), M_{interdependent} = 1.16 (SD = 0.33), t(58) = .073, p = .94; Positive behavior: M_independent = 6.46 (SD = 0.79), M_{interdependent} = 6.61 (SD = 0.63), t(58) = .83, p = .410; Negative emotions: M_independent = 1.39 (SD = 0.52), M_{interdependent} = 1.39 (SD = 0.44), t(58) = .027, p = .98; Positive emotions: M_independent = 5.36 (SD = 1.17), M_{interdependent} = 5.98 (SD = 0.71), t(58) = .48, p = .68).

Consistent with our hypotheses, following a negative experience, participants with an independent (vs. interdependent) self-construal were significantly more likely to express negative emotions and negative behaviors. Negative emotions partially mediated the effect of culture on behavioral response to the disappointing experience. In the next experiment we explored two factors which potentially underlie these cultural differences.

Experiment 2: The Mechanism Underlying the Effect

Independents and interdependents differ in dialectical thinking (Nisbett et al., 2001) and in emotional appraisals (Ellsworth & Scherer, 2003; Scherer, 1999). In this experiment we explored the mediating roles of these factors in the effect of self-construal on reactions to disappointing brand experiences.

Method

Participants were 120 undergraduate students (20% male, Mage = 24 years), who received extra course grade points for their participation. They were randomly assigned to one of two self-construal conditions. After completing the self-construal manipulation, participants read a description of a negative sports scenario, describing a soccer game in which the team played carelessly and eventually lost (Appendix B). They completed the same scales from experiment 1 to measure their emotional and behavioral reactions, and then completed measures of dialectical thinking and emotional appraisal.

To measure dialectical thinking, we used the 32-item Dialectical Self Scale (DSS; Spencer-Rodgers et al., 2004), which employs a seven-point scale (1 = strongly disagree; 7 = strongly agree), including items such as: “When I hear two sides of an argument, I often agree with both”.

To assess emotional appraisal, we used thirty questions, based on existing appraisal questionnaires (Frijda et al., 1989; Manstead et al., 1989; Smith & Ellsworth, 1985). It measured the following appraisals: intensity, pleasantness, unexpectedness from own perspective, unexpectedness from other’s perspective, inconsistency with own standards, inconsistency with other’s standards, benefit to self, benefit to others, own responsibility, other’s responsibility, impersonal control, bearable, self-esteem, fairness, interestingness, clearness, clearness outcome, end foretold, stand, suddenness, true expectedness, anticipated effort, personal importance, importance, modifiability, familiarity, knowledge, pleasure, suffering, and pain. The measure which employed a seven-point scale (1 = not at all; 7 = extremely, included items such as: “What was the extent to which the situation was pleasant to you?” (measuring the pleasantness appraisal) and “to what extent what the situation unexpected from your own perspective?” (measuring unexpectedness from one’s own perspective).

Results and Discussion

Self-construal significantly influenced negative behavioral intentions (M_independent = 3.47 (SD = 0.92), M_{interdependent} = 3.04 (SD = 1.12), t(114) = −2.26, p = .026). Participants in the independent prime condition were more likely than those in the interdependent condition to feel negative emotions (M_independent = 4.11 (SD = 1.11), M_{interdependent} = 3.67 (SD = 1.26), t(114) = −1.984, p = .05). This was true for an aggregated measure of all negative emotions, and specifically for anger (M_independent = 5.44 (SD = 1.14), M_{interdependent} = 4.93 (SD = 1.45), t(114) = −2.135, p = .035), anxiety (M_independent = 3.70 (SD = 2.29), M_{interdependent} = 2.43 (SD = 1.85), t(109.006) = −3.30, p = .001) and frustration (M_independent = 5.50 (SD = 1.22), M_{interdependent} = 4.69 (SD = 1.98), t(114) = −2.65, p = .009). Other emotions did not show significant effects: guilt (M_independent = 2.93 (SD = 1.89), M_{interdependent} = 2.50 (SD = 1.64), t(114) = −1.31, p = .19), fear (M_independent = 3.43 (SD = 2.04), M_{interdependent} = 2.93 (SD = 1.86), t(114) = −1.37, p = .17), shame (M_independent = 4.32 (SD = 1.87), M_{interdependent} = 4.12 (SD = 2.06), t(114) = −.56, p = .57), contempt (M_independent = 4.69 (SD = 1.86), M_{interdependent} = 4.36 (SD = 1.98), t(114) = −.92, p = .36), sadness (M_independent = 4.32 (SD = 1.78), M_{interdependent} = 4.24 (SD = 1.93), t(114) = −.25, p = .80), disgust (M_independent = 2.15 (SD = 1.51), M_{interdependent} = 2.31 (SD = 1.74), t(114) = .51, p = .61), and dejection (M_independent = 4.63 (SD = 1.82), M_{interdependent} = 4.26 (SD = 1.81), t(114) = −1.07, p = .29). Positive emotional reaction did not differ between conditions (M_independent = 2.45 (SD = 0.67), M_{interdependent} = 2.39 (SD = 0.66), t(114) = −.52, p = .61).

Participants in the interdependent condition were more comfortable with contradictions than those in the independent condition (DSS: M_independent = 4.44 (SD = 0.50), M_{interdependent} = 4.70 (SD = 0.53), t(114) = 2.64, p = .009). We also looked at the mediating role of DSS in the effect of self-construal on behavioral intentions. When behavioral intentions were regressed simultaneously on self-construal and DSS, the effect of self-construal on behavioral intentions remained significant (β = .19, t(113) = 2.031, p = .045), indicating that no mediation occurred.

Results revealed significant effects of self-construal on the following appraisal dimensions: pleasantness (M_independent = 1.48 (SD = 0.63), M_{interdependent} = 2.15 (SD = 1.20), t(86.19) = 3.79, p < .001), inconsistency with own standards (M_independent = 4.98 (SD = 1.28), M_{interdependent} = 4.37 (SD = 1.25), t(114) = −2.57, p = .012), inconsistency with others’ standards (M_independent = 4.81 (SD = 1.44), M_{interdependent} = 4.31 (SD = 1.22), t(114) = −2.02, p = .046), others’ responsibility (M_independent = 4.03 (SD = 1.83), M_{interdependent} = 3.29 (SD = 2.02), t(114) = −2.08, p = .040), being bearable (M_independent = 4.26 (SD = 1.46), M_{interdependent} = 4.81 (SD = 1.37), t(114) = 2.10, p = .038), and clearness (M_independent = 5.72 (SD = 1.12), M_{interdependent} = 5.12 (SD = 1.35), t(114) = −2.6, p = .010). Results regarding the other appraisals were insignifcant: intensity (M_independent = 4.96 (SD = 1.38), M_{interdependent} = 4.82 (SD = 1.44), t(114) = −.53, p = .60), unexpectedness from own perspective (M_independent = 3.70 (SD = 1.54), M_{interdependent} = 3.82 (SD = 1.23), t(114) = .47, p = .64), unexpectedness from other’s perspective (M_independent = 3.81 (SD = 1.51), M_{interdependent} = 3.62 (SD = 1.26), t(114) = −.73, p = .46), benefit to self (M_independent = 2.87 (SD = 1.55), M_{interdependent} = 2.65 (SD = 1.29), t(114) = −.84, p = .40), benefit to others (M_independent = 3.79 (SD = 1.37), M_{interdependent} = 3.91 (SD = 1.21), t(114) = .50, p = .62), own responsibility, (M_independent = 1.72 (SD = 1.28), M_{interdependent} = 2.08 (SD = 1.40), t(114) = −1.53, p = .128), impersonal control (M_independent = 4.01 (SD = 1.82), M_{interdependent} = 3.96 (SD = 1.82), t(114) = −.15, p = .88), self-esteem (M_independent = 3.29 (SD = 1.09), M_{interdependent} = 3.53 (SD = 1.07), t(114) = 1.20, p = .23), fairness (M_independent = 4.34 (SD = 1.56), M_{interdependent} = 4.29 (SD = 1.48), t(114) = −.18, p = .86), interestingness (M_independent = 4.60 (SD = 1.63), M_{interdependent} = 4.48 (SD = 1.50), t(114) = 3.790, p = .68), clearness outcome (M_independent = 4.98 (SD = 1.28), M_{interdependent} = 4.37 (SD = 1.25), t(114) = −.414, p = .012), end foretold (M_independent = 4.34 (SD = 1.51), M_{interdependent} = 4.13 (SD = 1.82), t(114) = −.67, p = .51), stand (M_independent = 5.39 (SD = 1.31), M_{interdependent} = 5.12 (SD = 1.41), t(114) = −1.09, p = .28), suddenness (M_independent = 3.48 (SD = 1.41), M_{interdependent} = 3.67 (SD = 1.41), t(114) = .72, p = .47), true expectedness (M_independent = 4.01 (SD = 1.45), M_{interdependent} = 3.81 (SD = 1.68), t(114) = −.71, p = .48), anticipated effort (M_independent = 3.32 (SD = 1.65), M_{interdependent} = 2.98 (SD = 1.78), t(114) = −1.08, p = .28), personal importance (M_independent = 3.36 (SD = 1.83), M_{interdependent} = 3.58 (SD = 2.05), t(114) = .62, p = .54), importance (M_independent = 3.81 (SD = 1.64), M_{interdependent} = 4.27 (SD = 1.80), t(114) = 1.45, p = .15), modifiability (M_independent = 3.24 (SD = 1.72), M_{interdependent} = 3.06 (SD = 1.63), t(114) = −.55, p = .58), familiarity (M_independent = 4.87 (SD = 1.59), M_{interdependent} = 4.55 (SD = 1.59), t(114) = −1.11, p = .27), knowledge (M_independent = 4.10 (SD = 1.60), M_{interdependent} = 4.24 (SD = 1.46), t(114) = .42, p = .67), pleasure (M_independent = 1.82 (SD = 1.25), M_{interdependent} = 1.63 (SD = .87), t(114) = −.94, p = .35), suffering (M_independent = 3.84 (SD = 1.70), M_{interdependent} = 3.84 (SD = 1.89), t(114) = .00, p = 1.00), and pain (M_independent = 3.15 (SD = 1.83), M_{interdependent} = 3.12 (SD = 1.97), t(114) = −.09, p = .92).

In line with our conceptualization, we conducted mediation analysis for the role of pleasantness appraisal in the effect of self-construal on behavioral intentions. First, it was found that self-construal (1: interdependent, 2: independent) was positively associated with intentions for negative behavior (c-path; B = .43, t(114) = 2.26, p < .05). It was also found that self-construal was negatively related to pleasantness appraisal (a-path; B = -.67, t(114) = 3.79, p < .001). Finally, the mediator pleasantness appraisal was negatively associated with intentions for negative behavior (b-path; B = -.23, t(114) = 2.33, p < .05). Because both the a-path and b-path were significant, the model met the criteria according to Baron and Kenny (1986). Mediation analysis was also tested using a bootstrapping method with bias-corrected confidence estimates (Preacher et al., 2007). In the present study, the 95% confidence interval of the indirect effects was obtained with 5000 bootstrap resamples (Preacher & Hayes, 2008). Results confirmed the mediating role of pleasantness appraisal in the relation between self-construal and intentions for negative behavior (B = .15; CI = 0.03, 0.33). The direct effect of self-construal on intentions for negative behavior was no longer significant (c′-path; B = .27, t(114) = 1.39, p > .05) when controlling for pleasantness appraisal, thus suggesting full mediation (Fig. 2). We also conducted the mediation analysis for the other emotional appraisals (which were significantly affected by self-construal); none of them was found to mediate the above effect.¹

Fig. 2

Indirect effect of self-construal on intentions for negative behavior through pleasantness (Experiment 2)

Consistently with experiment 1 and with our hypotheses, participants in the independent condition were more likely to express negative emotions and behaviors than those in the interdependent condition. Looking into the processes behind these findings showed that those in the interdependent condition were more comfortable with contradictions than those in the independent condition, yet no mediation was found. Emotional appraisal, and particularly pleasantness appraisal, was shown to play a mediating role in the effect of self-construal on behavioral intentions. Thus, findings from this experiment suggest that disappointing experiences are perceived as more unpleasant (i.e., cause more suffering) by independents (vs. interdependents), which in turn makes them more likely to express negative behavioral intentions.

Experiment 3: The Role of Prior Expectations

This experiment had two goals: (1) to replicate experiments 1 and 2 in a restaurant encounter context; and (2) to examine the role of prior expectations (hypothesis 2).

Method

Participants were 120 undergraduate students, who were randomly assigned to conditions of a 2 (independent vs. interdependent) × 2 (with vs. without expectations) between-subjects experiment. Participants performed the self-construal manipulation, and then read a scenario describing poor service at a restaurant, indicating either that the consumer had known the restaurant and expected a positive experience, or that the restaurant was new (Appendix C). Participants completed the emotional and behavioral intention scales used in experiment 1.

Results and Discussion

Overall (regardless of whether the consumer had or did not have prior expectations) self-construal significantly influenced negative behavioral intentions (Negative behavior: M_independent = 5.01, M_{interdependent} = 4.68, F(1,116) = 3.43, p_(1-tail) = .03). The interaction between expectations and self-construal on negative behavioral intentions was insignificant (F(1, 116) = 2.32, p = .13). Yet because we had directional hypotheses, we examined the simple effects (Rosnow and Rosenthal, 1995). As predicted, with prior expectations, the effect of self-construal on negative behavioral intentions was significant (M_independent = 5.20 (SD = 1.03), M_{interdependent} = 4.59 (SD = 0.90), F(1, 116) = 5.703, p = .017; Fig. 3), whereas in the absence of expectations, the effect was insignificant (M_independent = 4.83 (SD = 0.79), M_{interdependent} = 4.77 (SD = 1.21), F(1,116) = .055, p = .83).

Fig. 3

The effect of expectations and self-construal on behavioral intentions (experiment 3). (*) Error bars represent standard deviations

Overall (regardless of expectations), participants in the independent (vs. interdependent) condition were more likely to feel negative emotions (M_independent = 3.69 (SD = 0.85), M_{interdependent} = 3.38 (SD = 0.86), F(1, 116) = 3.88, p = .05). Positive emotions did not differ significantly (M_independent = 2.22 (SD = 0.56), M_{interdependent} = 2.24 (SD = 0.68), F(1, 116) = .035, p = .85). Interestingly, in this study, the main effect of self-construal was significant only on anger (M_{Anger_ind} = 6.16 (SD = 0.92), M_{Anger_int} = 5.31 (SD = 1.27), F(1, 116) = 18.00, p < .001). Other emotions did not show significant effects (Guilt: M_independent = 1.75 (SD = 1.09), M_{interdependent} = 1.70 (SD = 1.16), t(116) = −.24, p = .81; Anxiety: M_independent = 1.95 (SD = 1.44), M_{interdependent} = 1.81 (SD = 1.28), t(116) = −.54, p = .59; Fear: M_independent = 1.66 (SD = 1.03), M_{interdependent} = 1.46 (SD = .99), t(116) = −1.07, p = .284 Shame: M_independent = 2.48 (SD = 1.53), M_{interdependent} = 2.33 (SD = 1.53), t(116) = −.53, p = .59; Frustration: M_independent = 5.46 (SD = 1.28), M_{interdependent} = 5.316 (SD = 1.72), t(116) = −1.08, p = .28; Contempt: M_independent = 5.25 (SD = 1.27), M_{interdependent} = 4.81 (SD = 1.78), t(116) = −1.59, p = .11; Sadness: M_independent = 3.58 (SD = 1.77), M_{interdependent} = 3.21 (SD = 1.68), t(116) = −1.16, p = .25; Disgust: M_independent = 4.11 (SD = 1.87), M_{interdependent} = 3.95 (SD = 1.81), t(116) = −.49, p = .62; Dejection M_independent = 4.53 (SD = 1.69), M_{interdependent} = 4.10 (SD = 1.84), t(116) = −1.34, p = .18).

Furthermore, we examined the interaction between expectations and self-construal on anger. The interaction was significant (F(1, 116) = 4.33, p = .04) with an insignificant effect when there were no expectations (M_independent = 5.90 (SD = 1.03), M_{interdependent} = 5.47 (SD = 1.41),F(1,116) = 2.34, p = .13), and a significant effect when there were prior expectations (M_independent = 6.43 (SD = 0.73), M_{interdependent} = 5.17 (SD = 1.12), F (1,116) = 19.98, p < .001; Fig. 4). The interaction effects between expectations and self-construal on all other negative emotions were insignificant (Guilt: (F(1, 116) = .32, p = .57); Anxiety: (F(1, 116) = .89, p = .47); Fear: (F(1, 116) = .78, p = .54); Shame: (F(1, 116) = 1.92, p = .168); Frustration: (F(1, 116) = .229, p = .633); Contempt: (F(1, 116) = .015, p = .903); Sadness: (F(1, 116) = .011, p = .92); Disgust: (F(1, 116) = .087, p = .77); Dejection: (F(1, 116) = .042, p = .84)).

Fig. 4

The effect of expectations and self-construal on anger (experiment 3). (*) Error bars represent standard deviations

Finally, we tested the overall moderated mediation model relating self-construal (X), anger (M), intentions for negative behavior (Y) and the existence of prior expectations (W). We performed a 95% percent bias-corrected bootstrapping procedure using 5000 resamples to obtain confidence intervals (CIs; Hayes, 2009). The results demonstrated that the direct effect of self-construal on anger was significant (a-path; B = 2.1, t(116) = 3.31, p < .001). Anger also significantly influenced intentions for negative behavior (b-path; B = .36, t(116) = 4.76, p < .001). The moderating effect of the existence of prior expectations on the relationship between self-construal and anger was insignificant (B = -.30; CI = -0.56, 0.17). Since the range contain zero, there is no indication of moderated mediation. The entire analysis, therefore, partly supports our predictions.

Overall, this experiment replicated the effect of self-construal on responses to disappointing experiences in a restaurant context, and showed that the greater tendency of independents (vs. interdependents) to experience negative emotions and to express negative behavioral intentions is due to unmet expectations.

General Discussion

Following a disappointing experience, people with an independent self-construal have been shown to be more likely to express negative behavioral intentions than those with an interdependent self-construal. In the first experiment, we showed that negative emotions mediate the effect of self-construal on behavioral response to a disappointing experience. In the second experiment, two additional mediating factors were examined: dialectic thinking and emotional appraisal. As expected, the findings regarding acceptability of contradictions show that participants in the interdependent (vs. independent) condition were more comfortable with contradictions, but this factor did not mediate the effect of self-construal on responses to the disappointment. Instead, findings regarding the role of emotional appraisal show differences in the appraisal of pleasantness in the situation, which mediated the effect of self-construal on response to the disappointment. In the third experiment we replicated the effect in a restaurant context, showing that indeed, the effect occurs due to unmet expectations.

This research contributes to knowledge on people’s responses to brand performance, shedding light on the effect of self-construal on such responses (Escalas & Bettman, 2005; Swaminathan et al., 2007), and illuminates the effect of negative emotions evoked in brand experiences. Moreover, because independent and interdependent self-construal are chronically salient in individualist and collectivist cultures, respectively, our findings enhance the understanding of emotional processes across cultures. Past literature that examined the effect of self-construal on consumers’ responses to negative brand information has largely overlooked the role of emotions (e.g., Bechwati & Morrin, 2003; Swaminathan et al., 2007). Our findings demonstrate differences in emotional responses to disappointing brand experiences between independents and interdependents, consequently affecting their future behavior. As such, our research contributes to dealing with this deficiency by uncovering the emotional processes underlying the effects of disappointing brand experiences. More broadly, Russel and Yik (1996) posited that “cultural differences in the antecedents of emotion might suggest differences in how events are construed, and differences in how an event is construed would suggest differences in the emotional reaction to the event” (p. 167). The present research is a step toward understanding cultural differences in the antecedents and their consequences in emotional experiences.

Our research also offers practical implications. As disappointing brand experiences are inevitable (Mattila & Cranage, 2005), marketers should consider the consequences of such disappointments. Previous research suggests that disappointing brand experiences result in negative behavior (Aaker et al., 2004; Bizman & Yinon, 2002; Cheng et al., 2012; Cialdini et al., 1976; Keaveney, 1995; Swaminathan et al., 2007), and therefore marketers should both try to avoid such negative experiences and consider recovery strategies (Wong et al., 2016). Our findings imply that in interdependent contexts, investing in such recovery strategies may be less necessary. Instead, it might be more beneficial to invest in nurturing relationships with the consumers and enhancing self-brand connection. Moreover, our findings about the role of emotional response in the effect of disappointment on consumer responses suggest an additional recovery strategy in independent contexts that has not been considered thus far: an emotional reappraisal strategy. The emotional regulation literature suggests that reappraisal processes can change the emotion evoked in situations (Gross, 1998). Knowing that it is the emotional reaction that leads independents to respond more negatively implies that leading consumers to appraise the situation differently, might change their behavioral response. Along this line, for example, marketers can explain to consumers why the brand experience has failed in a way that reduces negative emotions. Such a shift in the emotional response might be an effective way to avoid negative behavioral responses. Another move that may assist as a recovery strategy in independent contexts might be to evoke interdependence in consumers. This might be done, for example, by using slogans that emphasize the similarity or connectedness between the consumer and the ingroup (e.g., “we are all in this together”).

A recent meta-analysis (Miao et al., 2019) indicates that service providers’ emotional intelligence is positively associated with the consumer’s perception of the service quality. In line with our research, it may be that this link is due to the enhanced ability of emotionally intelligent employees to reduce negative emotional reactions and in turn enhance positive evaluation (Golder et al., 2012). Furthermore, Miao and colleagues provide initial evidence that this link is contingent upon culture, as determined by dimensions other than independence-interdependence. Specifically, their meta-analysis reveals that the positive effect of emotional intelligence on service quality evaluations is stronger for cultures characterized by indulgence (vs. restraint), as well as in those characterized by short-term orientation (vs. long-term orientation). Unfortunately, the studies included in the meta-analysis were all conducted in collectivist countries (e.g., Taiwan, South Korea, India), which does not enable examination of the individualist-collectivist cultural dimension. Extending the examination of the importance of emotional intelligence may reveal that employees’ emotional intelligence is more crucial in individualist (independent) cultures.

Several limitations call for further research. The effect of self-construal on the responses to disappointing brand experiences may be due to norms relating to emotions, emotion regulation, dialectical thinking, and emotional appraisal. Not all of these factors were examined in our research. Further research is needed to uncover the role of these additional factors, particularly the extent to which identification with a brand may be involved in these processes. Second, independence and interdependence also vary in the link to ego-focused and other focused emotions (Markus & Kitayama, 1991). In our studies, we measured a wide variety of negative emotions. While we saw that participants in the self-construal conditions always differ in their ego-focused emotions (i.e., anger), we also saw differences in other emotions (e.g., shame). More research is needed to tap into the subtle differences in more specific emotions. Third, our examination used a priming manipulation as the operationalization for self-construal. Other operationalizations, such as cultural orientation based on nationality as well as measured self-construal, should be utilized in the future. In addition, we relied on a manipulation that has been validated in past studies, but did not include a measure of self-construal in our studies. Future studies should include such measures along with measures of the extent to which participants are focused on their own goals versus on the group goals, which may assist in reinforcing the proposed mechanism. Future studies should also add measures of additional factors that may be involved such as locus of control and power distance. Finally, future research may examine other cultural dimensions, such as masculinity versus femininity, uncertainty avoidance, and long-term orientation (Hofstede, 1983).

Appendices

Appendix A. The basketball game scenario (used in experiment 1)

Participants were asked to imagine themselves in the following situation:

Imagine that you are attending a basketball game of a team that you’ve been a fan of since you were a child. The game is a home game. At the end of the first half your favorite team is lagging behind by three points and the halftime score is 43–40. The third quarter begins with two turnovers of your team’s players that leads to two offense attacks of the opposing team. The coach of your team takes a time-out. After the time-out the team begins an attack that does not yield a basket, the opposing team embarks on an attack scoring a three-pointer, and now the scoreboard is 50–40. Your team continues to play in a sloppy manner and rarely scores, while the opposing team demonstrates full control on the court. At the end of the third quarter, the score is 67–46 in favor of the opposing team. The fourth quarter is a continuation of the third quarter and your team continues to lose balls and play irresponsibly, while the other team presents a delightful game to watch on both ends (offense and defense). There are six minutes left to the end of the game and the scoreboard stands at 79–52 (27 more points for the opposing team).

Appendix B. The soccer game scenario (used in experiment 2)

Participants were asked to imagine themselves in the following situation:

Imagine that you are at a soccer game of your favorite team, which you’ve been a fan of since childhood. The game is a home game. The opposing team is ranked at the top of the table and is considered the best team in the league. At the end of the first half, during which the soccer playing is of low quality, the scoreboard shows a tie, 0–0. The second half opens with several turnovers of your team, which leads the opposing team to perform several offense moves which gives them with two goals, and now the scoreboard shows a 2–0 lead for the opposing team. Your team’s coach feels that your defense is not functioning properly, and orders a change in tactics in order to strengthen the defense. Your team keeps playing recklessly, has no control over the game, and keeps turning over the ball due to inaccurate passes. The opposing team in the meantime keeps its absolute control over the game, and performs impressive offense moves which give the team another two goals and raises the score to 4–0. In order to avoid defeat, your team’s coach, who senses that his players are tired and aren’t giving their all, preforms another two switches in his defense. Despite these switches, your team keeps losing balls and playing in a sloppy manner, while the opposing team plays an enjoyable game both in offense and defense. When there are seven minutes left in the game, one of your team’s players violently fouls the opposing team’s best player. Due to this, the referee sends the offending player off and grants the opposing team a penalty kick, which results in another goal. The final whistle is heard and the opposing team wins 5–0, landing your team with a great loss.

Appendix C. The restaurant encounter scenario (used in experiment 3)

Participants were asked to imagine themselves in the following situation:

With expectations condition: Assume you and your friends are on your way to eat breakfast at your favorite restaurant, where you dine several times a month. [No expectation condition: Assume you and your friends are on your way to eat breakfast at a new restaurant that just opened at your hometown. You do not know the restaurant and it is you first visit there]. At your arrival, your hostess explains that due to the restaurant being full you can only sit in the north side of the restaurant, which is empty and distant from the rest of the diners. The hostess leads you to a small rectangular table which you estimate suits a maximum of two diners, and since you are four people and don’t want to be crowded when the food arrives you point to a larger table and ask to sit there. The hostess frowns and coldly answers “That’s a six-person table. You are four!”. Since the area you are sitting in is empty, you ask her why she cannot place you at the larger table. Yet she insists, without any courtesy or flexibility, that it is designated for six people. After ten minutes of waiting, you get up from your seat and wave twice to the waitress to come take the order. Your friends order “classic breakfast” and you order a green salad that comes with a bread basket. All the dishes come with a glass of orange juice and a plate of spreads, which arrive at your table seven minutes after ordering. The juice is warm, pale, and far from being “natural”, the spreads plate is meager and uninteresting. The breakfast your friends ordered and your green salad are fairly good. Your theory that the table would be too small for four people turns to be true, and when all the dishes arrived it was very crowded and your elbows kept touching. Close to the end of the meal you order an espresso which arrived after six minutes, but the wait isn’t worth it, since the espresso is burnt, and you don’t drink it. When the waitress asks if she can clear the table, you nod and reply that the espresso is undrinkable, yet the waitress doesn’t bother to ask if you want something else instead.