According to the temporal construal theory, distant future situations are construed on a higher level (i.e., using more abstract and central features) than near future situation [3] . Extending on this, researches on construal level theory suggest that a high versus low level of mental construal prompts people to think more abstractly [4] [5] . Specifically, they are likely to represent events in terms of general, superordinate, and decontextualized features rather than in terms of specific, subordinate, and contextualized features.

Prior researches on construal level theory proposed that a high level of mental construal motivates people to blame the features or characteristics of personality for something happened, while a low level of mental construal actuates people to blame the background or situations that things happened. According to these, consumers with high level of mental construal will blame the overseas brand for an overseas brand crisis. However, consumers with low level of mental construal will blame China background for an overseas brand crisis. Kyung et al. [6] also found that consumers faced with distant future situations are more likely to blame the Protagonist of an event. Körner and Volk proposed that consumers made more deontological judgments under concrete (vs. abstract) construal [7] . From the aboves, consumers will evaluate an overseas crisis more fairly and impartially under concrete (vs. abstract) construal. Therefore, consumers with high (vs. low) level of mental construal evaluate domestic brand more positively under the context of an overseas brand crisis.

According to social identity theory, when group status is threatened, highly committed group members respond with greater in-group favoritism [1] [8] . Petriglieri [2] proposed that there are two kinds of identity threat coping responses: identity-protection responses and identity restructuring responses. The former means to protect the threatened identity, and the latter to make it less of an object for potential harm. Similarity, White and Argo [9] find that when consumers experience a threat to aspect of their social identity (e.g., receive negative information regarding their national group), they sometimes avoid products associated with that identity and sometimes increase preference for identity-linked products. In conclusion, when consumers are highly identified with their country, they will defend the domestic brand under the context of an overseas brand crisis. However, consumers low in national identity will resist the identity-related brands or products under the context of an overseas brand crisis. In combine with the literatures on construal level theory above, we propose that national identity will moderates the effects of construal level on domestic brand evaluations. Hypotheses are as follows:

H1: To consumers with low national identity, they evaluate domestic brands more positively at higher (vs. lower) levels of construal under the context of overseas brand crises.

H2: The effects of construal levels on consumers’ evaluations of domestic brands will be attenuated for those with high national identity.

To examine H1 and H2, A 2 (construal level: high vs. low) × 2 (national identity: high vs. low) between-subjects experimental design was used. 60 undergraduate students (70% females, Mage = 18.7) from a business course at a large university in Southern China participated in exchange for course credit.

Firstly, participants were told that the survey was about university students’ mental health, and it includes three unrelated parts. The first part was to prime construal levels and measure participants’ national identities. According to Freitas et al. [10] , participants assigned to a high level of mental construal were directed to consider why they would engage in a health improvement activity, whereas participants assigned to a low level of mental construal were directed to consider how they would engage in the same activity. Then participants completed a six-item scale which measures national identities [11] (α = 0.819), all using 7-pointscales ranging from1 (strongly disagree) to 7 (strongly agree), This included items such as “When someone criticizes my country, it feels like a personal insult”, “When I talk about this country, I usually say ‘we’ rather than ‘they’”, “I am very interested in what others think about my country”.

The second part was a piece of real news about an overseas brand, named Johnson & Johnson which suffered from a crisis. In the overseas brand crisis condition, participants read a news report stating that while recalling defective products, the overseas brand, named Johnson & Johnson, excluded the Chinese market.

The third part was to rate the desirability of domestic brands stereotypes on a seven-point scales, including competence and morality dimensions. The measure consisted of seven stereotype attributes [12] (α = 0.769).three were assumed to refer to the moral realm (i.e. tolerant, modest) and four to the realm of competence (i.e. efficient, competitive). Participants were then asked to choose which T-shirts they preferred among three different ones. One is related to the in-groups, another to the out-groups and the third one is neutral.

According to Lee and Robbins [13] , the mean was used as a cutoff. Participants were then selected as a high-level national identity group member if they scored below the mean, and selected as a low-level group member if they scored the mean or higher.

According to the ANOVA analysis, the main effect of national identity was significant, F(1, 56) = 10.23, p < 0.01, η² = 0.154, participants with high national identity rated the domestic brand identities more favorably (M = 4.06, SD = 0.141) than participants with low national identity (M = 3.38, SD = 0.158). As predicted, the interaction of national identity and construal level on national brand identity desirability ratings was significant (F(1, 56) = 5.63, p < 0.04, η² = 0.09) (see Figure 1). Planned contrasts revealed that participants with low level of national identity evaluate the domestic brand identities more positively when they are assigned to the high level of construal (M = 3.86, SD = 0.588) than when assigned to the low level of construal (M = 2.89, SD = 0.97), F(1, 25) = 10.53, p < 0.01, η² = 0.296). Thus, H1 was supported. However, when with high level of national identity, participants assigned to the high level of construal (M = 4.04, SD = 0.91) and to the low level of construal (M = 4.07, SD = 0.76) perceived the same identities of the domestic brands, F(1, 31) = 0.013, p > 0.1. Thus, H2 was supported.

Furthermore, we examined the interactions effects on competence and moral dimensions of domestic brand identities. An ANOVA on the competence dimensions of domestic brand identities yielded a significant interaction, F(1, 56) = 4.74, p < 0.04, η² = 0.078 (see Figure 2). However, the interaction effects on the morality dimensions of domestic brand identities were not significant, F(1, 56) = 2.23, p > 0.1 (see Figure 3).

A logistic regression analysis indicated that the evaluations of competence dimensions of domestic brand can affect consumers’ choices T-shirts (χ² = 7.05, p < 0.01). In addition, national identity had a significant effect on the choice of T-shirts. Individuals with high level of national identity were more likely to choose the T-shirts related to the in-groups (χ² = 4.49, p < 0.05; M_{low national identity} = 33.3%, M_{high national identity} = 60.6%).