The social identity perspective [2, 4] holds that individuals can respond to a given situation in ways that are consistent with either their personal identity or one of the many possible social identities, e.g. national identity [5]. The motivational core of this theory emphasizes that individuals strive to achieve or maintain a positive social identity, which depends on favorable comparisons between their in-group and some relevant out-group. A social identity threat arises when the in-group is not perceived as positively differentiated or distinct from the relevant out-group. According to Tajfel and Turner [2], the dominant strategy for achieving positive social identity is social mobility, in which members of a threatened group can actually leave or psychologically dissociate themselves from their in-group. Psychological escape from one’s in-group dissociation can take many forms, including decreased similarity to the in-group [6, 7]. Researchers have also demonstrated that individuals who are strongly committed to a group tend to cope with social identity threats by adopting association with the group, that is, by increasing homogeneity with their in-group [8-10]. Individuals who strongly identify with an in-group might display out-group derogation under a social identity threat, which can be considered an extreme form of dissociation from the out-group [11, 12].

Group identity maintenance processes are closely related to personal identity processes [13, 14], and therefore, association/dissociation with social groups may be used strategically as a buffer against the threat to personal identity [1]. Mussweiler et al. [15] stated that personal and social identity are interrelated aspects of a single self-concept that comprises the general sense of self-worth. Based on this claim, it can be argued that social identity threat leads to similar reactions as personal identity threat, as both threaten the general sense of self-worth. Thus, the social identity theory may serve as a unified theoretical perspective in explaining group identity maintenance processes (association/dissociation) caused by a social identity threat and a personal identity threat. This assumption is supported by Cialdini et al. [16, 17], who demonstrated that, under a personal identity threat, individuals enhanced the asserted quality of their home university (i.e., associated with an in-group) and devalued the asserted quality of a rival university (i.e., dissociated from an out-group). Cialdini et al.’s research and analysis of literature suggest that associative or dissociative responses to social and personal identity threats can be similar and that dissociation can be applied not only to in-groups but also to out-groups.

Furthermore, White et al. [18] found that the type of self-construal determined whether individuals coped with social identity threats by adopting associative or dissociative responses. In light of this, the self-construal theory was applied to investigate the factors determining the response to an identity threat.

Self-construal is the degree to which people see themselves as separate from or connected with others [3]. Initially, two types of self-construal were distinguished: independent and interdependent. Individuals with a more independent self-construal (independents) see themselves as autonomous, unique, and distinct from the group [19]. In contrast, those with a more interdependent self-construal (interdependents) view themselves as connected to others, such that the self is defined by important roles, relationships, and group memberships. Further, three separate lines of research [20-22] have identified two types of interdependent self-construal, that is, relational-interdependent self-construal (close relationships are included in the self) and collective-interdependent self-construal (group memberships are included in the self).

The self-construal theory is distinct from the construal level theory that describes the relationship between psychological distance and the extent to which people's thinking (representing objects mentally) is concrete or abstract [ 23 ]. The general idea of the construal level theory is that individuals tend to think about close objects in more concrete ways and remote items in more abstract ways [ 24 ]. In contrast, the self-construal theory focuses on describing the extent to which social objects (other people and groups) are incorporated into the structure of the self [ 3,20].

Self-construal moderates the responses to personal identity threats [25, 26] and social identity threats [18]. White, Argo, and Sengupta [18] found that independents demonstrate a dissociative response when their social identity is threatened, whereas interdependents exhibit an associative response. Their studies demonstrate that different types of motivations are embedded in different responses to social identity threats for independents and interdependents. While interdependents access a repertoire of social identities in response to social identity threats to fulfill their needs for belongingness, independents are motivated to restore positive self-worth. These findings suggest that, under a social identity threat, independents engage in dissociation to protect their personal identity, manifested in self-worth concerns. Similar reactions to social identity threats are observed when personal identity is threatened. Vohs and Heatherton [26] found that under personal identity threats, interdependents behaved in a way that they were perceived as more likable, which can be interpreted as associative reactions toward other people and social groups. Meanwhile, under a personal identity threat, independents exhibit increased dissociative tendencies, as they behave in a less friendly way toward other people.

Thus, it seems that the type of self-construal determines the type of response to an identity threat rather than the type of threat. The self-construal, which determines an individual's relationship with social groups, may influence how an individual will respond to an identity threat. Individuals for whom belonging to a social group is vital (e.g., interdependents) may cope with an identity threat through greater association with their in-groups, which will give them a sense of belonging and security [18]. Such a sense of belonging and security will not be provided by out-groups, from which the interdependents will distance themselves as they associate with their in-group. Individuals who do not value membership of a social group (i.e., independents) should cope with identity threats by dissociating from social groups and seeking a sense of self-worth and security outside them (e.g., by strengthening personal identity).

Based on the abovementioned findings on responses to social and personal identity threats and the role of self-construal in these processes, the following hypotheses were proposed:

H1: For independents, a social and personal identity threats lead to dissociation from in-groups (H1a) and out-groups (H1b).

H2: For interdependents, social and personal identity threats lead to association with in-groups (H2a) and dissociation with out-groups (H2b).

In a study by White et al. [18], the processes of association or dissociation were expressed as changes in product preferences, symbolizing group membership. Hence, consumer psychology seems to be the perfect platform to capture the subtle processes of association or dissociation with social groups. This approach is justified as individuals use commercial brands to strengthen their sense of belonging to social groups [27], and incorporate brands into their self-concept, which is referred to as self-brand connection [28]. Individuals reported higher self-brand connections to brands symbolizing their in-group whereas lower connections to brands symbolizing out-groups.

Many studies have addressed the impact of social and personal identity threats on consumer preferences, thus showing the compensatory function of consumption. Consumers dissociate from unfavorable social groups to avoid social identity threats [ 27-29]. For instance, Spangenberg and Angle [30] described that consumers avoid previously held preferences when these become threats by dissociative group use. Further, consumers have been shown to respond to social identity threats by dissociation via avoiding identity-linked brands or by association via approaching them [18,31,32]. White and Argo [31] demonstrated this effect in their study. The authors found that when females who had low in-group identification received negative information about their gender identity (social identity threats), they avoided products that are symbolically associated with their gender (dissociated from in-group). In turn, females with high in-group identification were more likely to prefer the products associated with their gender (associated with in-group). In another study, White, Stackhouse, and Argo [32] found that, under social identity threats, consumers with low in-group identification avoided products reflecting their in-group identity more when public self-awareness was high as opposed to low, particularly in the presence of an in-group audience. Overall, when social identity is threatened, consumers with high rather than low public self-awareness are more likely to associate with their in-group through consumers' choices.

Similarly, studies indicate the activation of consumer association and dissociation in response to the personal identity threat. For instance, when consumers encountered a threat to the self-brand association, they reacted by association with the brand [ 33 ]. In comparison, consumers experiencing personal identity threats avoid low-status goods to prevent further damage to their self-worth [ 34 ]. However, only a few studies have examined the impact of personal identity threats on consumer association and dissociation processes with group identity. Examples of these studies are those of the Cialdini team that is described above [ 16,17].

Social and personal identity threats have been linked to increased anxiety along with intense physiological arousal [35-38]. Some studies have suggested that anxiety could operate as a mediator between threats to particular aspects of identity and behavioral changes [39, 40]. When identity is threatened, individuals experience anxiety and are motivated to reduce this unpleasant emotional state [41]. Recently Ma and Hmielowski [42] demonstrated that identity threat increased anxiety, leading to avoidance behavior. Another recent study showed that anxiety affects association and dissociation expressed in consumers' preferences and that self-construal moderates the associative reactions to anxiety [43]. However, the role of anxiety resulting from social and personal identity threats in the process of associative/dissociative responses has not yet been studied. It has been proposed that one of the primary motivations for coping with identity threats may be the reduction of unpleasant anxiety arousal. Thus, Hypothesis 3 was proposed:

H3. Social and personal identity threats affect the strength of association/dissociation processes through anxiety arousal.

We planned the sample size based on previous studies using similar study designs [18], where between 25 and 30 participants were assigned to each condition. We planned to examine between 30 and 35 participants per condition. Participants were 204 undergraduate women, aged between 18 and 25 years (M = 21.20, SD = 1.96). Undergraduate women participating in the study were recruited with advertisements about the study placed in many buildings of different departments of the three universities in Poznan, Poland. The invitation to participate in the study included information that, in compensation for taking part in the study, the participants would receive a cinema voucher. Volunteers were admitted into the study on the condition that they did not meet any exclusion criteria: significant health problems, use of medications, prior diagnosis of cardiovascular disease, or hypertension. Participants were instructed to avoid eating for at least one hour before the experiment and refrain from physical exercise. The participants were also instructed to avoid caffeine, nicotine, alcohol, or non-prescription drugs for at least two hours before the experiment. Five of them disbelieved the cover story for the identity threat procedure during debriefing and were excluded from further analyses. The study had a 3 (identity threat: social vs. personal vs. no threat) × 2 (self-construal: independent vs. interdependent) between-subjects experimental design. The Institutional Ethics Committee approved the study, and informed consent was obtained from all participants involved in the study.

Upon arrival in the laboratory, participants provided written informed consent, and were told that they would participate in two unrelated studies. Participants were randomly assigned to one of the six study conditions. Experimenters did not know the research hypotheses and did not know to which study condition participants were assigned. The biosensors were attached, and the experiment began with a 5-minute baseline, where the last 90 seconds were taken as a physiological pretest. The “first” study began with answers to the Oxford Test of Professional Competence (OTPC). Then, the manipulation of self-construal using the SST was conducted. Next, participants in experimental conditions received negative feedback regarding their individual (personal identity threats) or group performance (social identity threats) in the OTCP (see Appendix). In the no threat condition, participants received a neutral description of the OTPC. Afterwards participants waited 90 seconds (physiological post-test) for the beginning of the “second” study, in which they were asked to indicate their evaluations of four different pairs of brands. The logos of two brands (brands A and B) in each pair were presented simultaneously, and the answers on the preferences scale and the Self-Brand Connection Scale were given alternately; participants alternately evaluated brands A and B. Next, participants completed a scale-measured retrospective evaluation of threat and evaluated their familiarity with all eight brands used in the study. Finally, participants completed a suspicion probe and were debriefed. The experiment was conducted with the e-Prime 2.0 Professional Edition environment.

Ratings of specific brands were not included if the participants did not know the brand; that is if they rated “1” on the brand familiarity scale. To calculate preference for feminine, masculine, and unisex brands and connection between self and those brands, only familiar brands were used (familiarity greater than 1). There were no missing data for determining brand preferences and self-brand connections. Five participants did not know either of the masculine brands and thus I did not calculate association/dissociation with/from the out-group index for them. Data were analyzed in an association/dissociation process in three steps. First, 3 (identity threat: social vs. personal vs. no threat) × 2 (self-construal: independent vs. interdependent) two-way ANOVAs were run separately for four dependent variables (preference for feminine and masculine brands, self-brand connections with feminine and masculine brands). Next, a simple effects analysis was run to test the effect of identity threat separately for two types of self-construal for significant interactions. In the end, comparisons to test differences between the no-threat condition and the personal and social identity threat conditions separately for independents and interdependents were held. Compared with the no-threat condition, significantly higher or lower levels of dependent variables for the personal or social identity threat conditions were indicators of association/dissociation processes. Repeated measures ANOVAs were run for eight physiological indicators of anxiety arousal with one between-subject factor (separately: social identity threat vs. no threat; personal identity threat vs. no threat), and one within-subject factor (baseline and reactivity).

There are some limitations to this study. First, a student sample was used; therefore, the findings may not be generalizable to less-educated peers or other age cohorts. Second, the participants were only women; thus, it is necessary to examine whether similar pathways and processes of association/dissociation can be observed among men. Third, future studies can consider social categories other than sex (e.g., nationality). Fourth, this study is also limited because it was used to test a self-construal priming technique, and future studies could replicate the present results by including individual differences in self-construal in the research design. Moreover, future research may investigate whether the collective-relational self-construal, which was not manipulated in this study, can moderate association/dissociation processes. Fifth, the study results are limited to one type of social identity threat, that is, a threat to the positive self-group image. Future research could consider other forms of social identity threat, such as the threat to the distinctiveness of one's own group [53-56]. Overall, additional studies using different manipulations should be done to support the findings.