C. Yoo, D MacInnis /Journal of Business Research 58(2005)1397-1406 and Ab can be positive when these negative feelings from the types of cues in ads are not mediated through an evaluative ad are relevant to the commercial,s goals. Or the sad and system, but are direct in their effect. Aad, in turn, strongly sorrowful may provide viewers warmth feelings in this predicts brand attitudes. While attitudes are not belief-based, context they do seem to be directly affected by evaluative aspects of the ad that are not mediated through aad 5.5. Comparison of fits by ad execution condition When the ad execution is informational. the creation of positive brand attitudes appears to lie in creating favorable Because our idea is that ad execution format moderates evaluative thoughts and forming strong beliefs. Interest the brand attitude formation process, we ran two separate ingly, while each influences brand attitudes, favorable models--one for the emotional ad execution and one evaluative thoughts do not influence brand attitudes througl the informational ad execution, with paths corresponding the mediational influence of beliefs. Aad regarded as to those shown in Fig. 1. For each execution, we tested the meaningful and credible also produce more positive and paths shown in Fig. la and b. That is, we tested the extent fewer negative feelings. Such ads also influence ad attitudes to which the data from the emotional ad execution through the mediational influence of feelings. Moreover, condition fit the brand attitude formation process specified Aad mediates the effect of feelings on brand attitude in Fig. la and b. We also tested the extent to which the Interestingly, the brand attitude formation process for ads data from the informational execution condition fit the using informational appeals does not appear to represent the brand attitude formation process specified in Fig. la and b. viewer as a purely rational and emotionless information When we conducted the test, we included only significant processor. While such ads are designed to appeal initially to ths in Fig. 2 the viewer's rationality and while brand attitudes are Overall fit for emotional ad execution mod the affected by beliefs, feelings also play a dominant role in data from the emotional ad execution (x(46)=42.04, p= the brand attitude formation process. Likewise, the brand 162: GFI=952; AGFI=930)is better than with the data attitude formation process for ads using an emotional appea from the informational ad execution condition (x(49) does not appear to be purely affect-driven More cognitive 81.82, p=030: GFI=868: AGFI. 824). When the data driven outcomes like evaluative thoughts play an important from informational ad execution condition was applied to role in the ad effectiveness of emotional appeals, despite emotional ad execution model, the path from credibility to these ads'initial appeals to the viewer's emotions beliefs becomes insignificant. Also, overall fit for informa If these results are generalizable, a key managerial take- tional ad execution model with the data from the informa- away is that if the ad execution is emotional, it is critical that tional ad execution (x(50)=65.32, p=088; GFI=920; it be exceptionally emotionally evocative, as responses later AGFI. 878)is better than with the data from the emotional in the ad persuasion process are contingent on the nature and ad execution condition(x(50)=62.84, P=072: GFI= 878; level of emotional responses generated. If, on the other AGFI. 805). When the data from informational ad execu- hand, the execution is designed to be informational, tion condition was applied to emotional ad execution model, critical that the ad be regarded as credible and meaningful the path from negative feelings to Aad becomes insignif- From a pretesting standpoint then, our research identifies icant. These results indicate that we need to run two separate that different criteria are primary to ad pretesting, depending models by ad execution conditions on whether the ad is emotional or informational Our results also suggest that while certain variables may be critical assessment tools. one should not assume that 6. Conclusions feeling ads are devoid of cognitive influences on persuasion or vice versa. Feelings- and credibility-related responses The results of this study indicate that the brand attitude were related to Aad and ab regardless of whether the ad was ormation process depends on whether ads use an emotional emotional or informational. Hence, while certain variables or information executional format. When the commercial may be critical to diagnosing the likely success of emotional uses an emotional format, brand attitudes are apparently and informational ads, success of both ad format types is driven through feeling responses. Positive feelings enhance contingent on providing favorable feelings and evaluations evaluations of the ads credibility, while most negative of the ad as credible. clings result in negative evaluations of the ad and brand Interestingly, however, some negative feelings, which 6.1. Limitations and future research directions appear relevant to the communication, exhibit a positive influence on ad evaluations. Because the arousal of positiv Although this research had some control over ad-related feelings and appropriate negative feeling makes the ad more execution variables, generalizations from this study are ppealing, those feelings positively influence ad attitudes limited given the single ad, single product category used in through the mediational effect of evaluative thoughts. the study. Exposure is also artificial as subjects were Negative feelings unrelated to the message, however, exert instructed to watch ads as opposed to having ads placed a direct negative effect on ad attitudes. Apparently some in a normal exposure context. Moreover, the emotional vs.and Ab can be positive when these negative feelings from the ad are relevant to the commercial’s goals. Or the sad and sorrowful may provide viewers warmth feelings in this context. 5.5. Comparison of fits by ad execution condition Because our idea is that ad execution format moderates the brand attitude formation process, we ran two separate models—one for the emotional ad execution and one for the informational ad execution, with paths corresponding to those shown in Fig. 1. For each execution, we tested the paths shown in Fig. 1a and b. That is, we tested the extent to which the data from the emotional ad execution condition fit the brand attitude formation process specified in Fig. 1a and b. We also tested the extent to which the data from the informational execution condition fit the brand attitude formation process specified in Fig. 1a and b. When we conducted the test, we included only significant paths in Fig. 2. Overall fit for emotional ad execution model with the data from the emotional ad execution (v2 (46) = 42.04, p = 162; GFI =.952; AGFI =.930) is better than with the data from the informational ad execution condition (v2 (49) = 81.82, p =.030; GFI =.868; AGFI =.824). When the data from informational ad execution condition was applied to emotional ad execution model, the path from credibility to beliefs becomes insignificant. Also, overall fit for informa￾tional ad execution model with the data from the informa￾tional ad execution (v2 (50) = 65.32, p =.088; GFI =.920; AGFI = .878) is better than with the data from the emotional ad execution condition (v2 (50) = 62.84, p =.072; GFI =.878; AGFI = .805). When the data from informational ad execu￾tion condition was applied to emotional ad execution model, the path from negative feelings to Aad becomes insignif￾icant. These results indicate that we need to run two separate models by ad execution conditions. 6. Conclusions The results of this study indicate that the brand attitude formation process depends on whether ads use an emotional or information executional format. When the commercial uses an emotional format, brand attitudes are apparently driven through feeling responses. Positive feelings enhance evaluations of the ad’s credibility, while most negative feelings result in negative evaluations of the ad and brand. Interestingly, however, some negative feelings, which appear relevant to the communication, exhibit a positive influence on ad evaluations. Because the arousal of positive feelings and appropriate negative feeling makes the ad more appealing, those feelings positively influence ad attitudes through the mediational effect of evaluative thoughts. Negative feelings unrelated to the message, however, exert a direct negative effect on ad attitudes. Apparently some types of cues in ads are not mediated through an evaluative system, but are direct in their effect. Aad, in turn, strongly predicts brand attitudes. While attitudes are not belief-based, they do seem to be directly affected by evaluative aspects of the ad that are not mediated through Aad. When the ad execution is informational, the creation of positive brand attitudes appears to lie in creating favorable evaluative thoughts and forming strong beliefs. Interest￾ingly, while each influences brand attitudes, favorable evaluative thoughts do not influence brand attitudes through the mediational influence of beliefs. Aad regarded as meaningful and credible also produce more positive and fewer negative feelings. Such ads also influence ad attitudes through the mediational influence of feelings. Moreover, Aad mediates the effect of feelings on brand attitudes. Interestingly, the brand attitude formation process for ads using informational appeals does not appear to represent the viewer as a purely rational and emotionless information processor. While such ads are designed to appeal initially to the viewer’s rationality and while brand attitudes are affected by beliefs, feelings also play a dominant role in the brand attitude formation process. Likewise, the brand attitude formation process for ads using an emotional appeal does not appear to be purely affect-driven. More cognitive￾driven outcomes like evaluative thoughts play an important role in the ad effectiveness of emotional appeals, despite these ads’ initial appeals to the viewer’s emotions. If these results are generalizable, a key managerial take￾away is that if the ad execution is emotional, it is critical that it be exceptionally emotionally evocative, as responses later in the ad persuasion process are contingent on the nature and level of emotional responses generated. If, on the other hand, the execution is designed to be informational, it is critical that the ad be regarded as credible and meaningful. From a pretesting standpoint then, our research identifies that different criteria are primary to ad pretesting, depending on whether the ad is emotional or informational. Our results also suggest that while certain variables may be critical assessment tools, one should not assume that feeling ads are devoid of cognitive influences on persuasion or vice versa. Feelings- and credibility-related responses were related to Aad and Ab regardless of whether the ad was emotional or informational. Hence, while certain variables may be critical to diagnosing the likely success of emotional and informational ads, success of both ad format types is contingent on providing favorable feelings and evaluations of the ad as credible. 6.1. Limitations and future research directions Although this research had some control over ad-related execution variables, generalizations from this study are limited given the single ad, single product category used in the study. Exposure is also artificial as subjects were instructed to watch ads as opposed to having ads placed in a normal exposure context. Moreover, the emotional vs. 1404 C. Yoo, D. MacInnis / Journal of Business Research 58 (2005) 1397 – 1406