Literature review

  While there are many conflicting research results regarding the effect of provocative advertising, there seems to be a unanimous view that provocative ads attract people’s attention (e.g., Bello, Etzel, & Pits, 1983; Dahl et al., 2003) and bring about better recall of the ad and the brand (Vezina & Paul, 1994) in both academic and business communities. As many researchers have ascertained that attraction to an ad precedes the elaborative processing of the ad (Celsi & Olson, 1988; Huhmann, 2007; MacInnis, Moorman, & Jaworski, 1991), we hypothesize that people would have more elaboration on provocative ads than non-provocative ads. In fact, Huhmann and Mott-Stenerson (2008) found that controversial advertising generated more elaborative processing than non-controversial advertising. Because there is little reason to believe that this pattern of elaborative processing would be different whether or not there is a social cause embedded in provocative advertising, we hypothesize higher elaboration on provocative advertising than on non-provocative advertising regardless of the embedment of a social cause.

  H1: People will have more elaboration on provocative advertising than on non-provocative advertising

  Previous research has found that involvement, in general, increased elaborative processing (Celsi & Olson, 1988; Maheswaran & Chaiken, 1991). However, as Huhmann and Mott-Stenerson’s (2008) study showed, product involvement in the context of provocative advertising did not moderate the relationship between provocation in advertising and elaboration. We reason that although involvement, in general, is a potent antecedent to elaboration, product involvement specifically is not significant enough to moderate the relationship between provocation in advertising and elaboration. In this study, we propose that higher involvement with the social issue addressed in the ad (Child welfare in this study), instead of product involvement, would bring about more elaboration. We argue that, however, if a subject is exposed to provocative ads, the increment of elaboration caused by the level of issue involvement would be miniscule because they are already prone to elaborate the ad due to the highly provocative nature. Therefore, we hypothesize that people who are exposed to provocative ads would be more likely to elaborate on the ads regardless of the level of issue involvement, but people who are exposed to the non-provocative ads would be more likely to elaborate on the ads when they are highly involved with the social issue addressed in the ad.

 H2: There will be an interaction effect between provocation in advertising and the issue involvement on the elaboration level. That is, in provocative advertising, there will be no difference in the elaboration level between people highly involved and people less involved with the social issue addressed in the ad. However, if people are exposed to non-provocative advertising, people highly involved with the social issue addressed in the ad will be more likely to elaborate on the ad than people less involved.

  The persuasion knowledge model proposes that if people are not aware of the persuasive intent of a persuasive agent, they are more likely to be persuaded than those who are aware of the persuasive intent (Friestad & Wright, 1994; Kirmani & Zhu, 2007; Yoo, 2009). Campbell and Kirmani (2000) found that if an ulterior motive for a persuasive agent’s act was highly accessible, target people were more likely to use their persuasion knowledge. Therefore, in the context of advertising, an advertiser should make its ulterior persuasive intent less accessible to consumers to keep them from using their persuasion knowledge (Cowley & Barron, 2008; Nelson, Wood, & Paek, 2009; Van Reijmersdal, Neijens, & Smit, 2005; Wei, Fischer, & Main, 2008). However, it is not always possible because an advertiser’s intent in some advertising appeals is inevitably more apparent than that in other advertising appeals. For example, if an ad keeps drumming up people on why they need to buy the product and where to get it, it is hard not to notice what the ad is trying to accomplish. By a similar reasoning, the persuasive intent of provocative advertising would be more prominent than that of non-provocative adverting because it is perceived that advertisers go the extra mile to attract people’s attention.

 This perception, however, could be mitigated by the introduction of some factors into the ad, such as a social cause (Pope et al., 2004). We propose that the activated persuasion knowledge in provocative ads would be decreased with the embedment of a social cause because of its desirable nature. Of course, even when consumers are exposed to non-provocative ads which address a social cause, it is possible that their inferential thinking might be torn between the mission of a social cause and the commercial motive of an advertiser and the commercial motive of an advertise why the company advertises. However, the difference between the activation of persuasion knowledge between the ad which addresses a social cause and the ad which does not address a social cause will be greater when the persuasive motive of an advertiser is apparently obvious as in the provocative advertising appeal.

 H3: There will be an interaction effect between provocation in advertising and the embedment of a social cause on the activation of persuasion knowledge. That is, in provocative advertising, people who are exposed to the ad which does not address a social cause will draw on their persuasion knowledge more than those who are exposed to the ad which addresses a social cause. On the other hand, in non-provocative advertising, there will be no difference in the level of activation of persuasion knowledge between those who are exposed to the ad which addresses a social cause and those who are exposed to the ad which does not address a social cause.

  We hypothesized that people would have more elaboration on provocative advertising than non-provocative advertising in hypothesis 1. However, higher elaboration alone cannot explain why some provocative ads are effective and others not. That is, unless one knows the ratio of positive or negative thoughts among all thoughts people listed after they are exposed to the ads, the total number of thoughts alone reveals little about the effectiveness of provocative ads. In hypothesis 4, we reason that people who are exposed to provocative ads which address a social cause would have more positive thoughts than those who are exposed to provocative ads which do not address a social cause. However, in non-provocative advertising, the embedment of a social cause might not be a significant factor on the ratio of negative thoughts to the total number of thoughts because of the less involving nature of non-provocative ads.

 H4: There will be an interaction effect between provocation in advertising and the embedment of a social cause on the ratio of negative thoughts to the total number of thoughts. That is, when people are exposed to provocative advertising, they will list a higher ratio of negative thoughts if the ad does not address a social cause. On the other hand, when people are exposed to non-provocative advertising, there will be no difference in the ratio of negative thoughts between those who are exposed to the ad which addresses a social cause and those who are exposed to the ad which does not address a social cause.

  The effect of provocation in advertising on attitude measures (Attitude toward the ad, attitude toward the brand and purchase intention) has been found to be negative in general (Sabri & Obermiller, 2011; Vezina & Paul, 1994). Vezina and Paul (1997) found that provocative advertising had a negative effect on the attitude toward the brand and purchase intention. Chan, Li, Diehl and Terlutter (2007) also found that if people perceived an ad as offensive, they were less likely to purchase the product and this phenomenon was more evident among Chinese consumers. Although the effect of provocative advertising on the attitude toward the ad was positive in some cases, it could be considered as an anomaly (Pope et al., 2004).

  We suspect that the effect of provocative advertising has been predominantly negative because researchers have been more interested in the negative side of provocative advertising (Andersson, Hedelin, Nilsson, & Welander, 2004; Bello et al., 1983). De Pelsmacker & Van Den Bergh (1996) developed a scale of 15 items for the measurement of provocation in advertising. Their study offered a valuable insight because they found not only negative attributes but also positive attributes of provocation in advertising. They found that provocative ads could be arresting, curiosity- provoking, moving and daring. Therefore, we can assume that people would react differently to different types of provocative advertising in terms of attitude. As we hypothesized in hypothesis 4, if the ratio of negative thoughts amplifies when the provocative ad does not address a social cause, we also could expect that the negative attitude effect of provocative advertising is more likely to be visible when it does not address a social cause.

 H5a: There will be an interaction effect between provocation in adver-tising and the embedment of a social cause on the attitude toward the ad. That is, when people are exposed to provocative advertising, they will react more positively to the ad if the ad addresses a social cause. On the other hand, when people are exposed to non-provocative advertising, there will be no difference in the attitude toward the ad between those who are exposed to the ad which addresses a social cause and those who are exposed to the ad which does not address a social cause.

  H5b: There will be an interaction effect between provocation in advertising and the embedment of a social cause on the attitude toward the brand. That is, when people are exposed to provocative advertising, they will react more positively to the brand if the ad addresses a social cause. On the other hand, when people are exposed to non-provocative advertising, there will be no difference in the attitude toward the brand between those who are exposed to the ad which addresses a social cause and those who are exposed to the ad which does not address a social cause.

  H5c: There will be an interaction effect between provocation in advertising and the embedment of a social cause on purchase intention. That is, when people are exposed to provocative advertising, they will have a higher purchase intention if the ad addresses a social cause. On the other hand, when people are exposed to non-provocative advertising, there will be no difference in purchase intention between those who are exposed to the ad which addresses a social cause and those who are exposed to the ad which does not address a social cause.

Method

 Variables of the Study

 In the current study, there are two independent variables, six dependent variables, and a moderating variable. The first independent variable is the level of provocation in advertising. In general, the lack of agreement on the conceptual definition of provocation in previous studies has led to con-fusion and insufficient attention to the proper empirical application of the concept. Some researchers (e.g., Bello et al., 1983) checked the manipulation of provocation in advertising with a sexual or erotic nature of the ad which cannot be identified with provocation in advertising by itself. While the sexual nature of advertising could be a factor which contributes to the perceived level of provocation, sexual and provocative appeals should be considered differently (Vezina & Paul, 1997). In other words, one should distinguish a possible factor of provocation (e.g., sex, fear) from the perceived provocation that a consumer feels after he or she is exposed to an ad.

 Pope et al. (2004) checked the manipulation of provocation using De Pelsmacker and Van Den Bergh’s (1996) 15 items. Although these two studies reported bigger than .70 Cronbach’s alphas, some items in the list were too specific to apply in many types of provocative advertising. For example, the item “sexually defiant” might not be applicable to a provocative ad which is not sexually oriented. Some researchers used more general items to measure provocation in advertising (Dahl et al., 2003; Huhmann & Mott-Stenerson, 2008). However, these researchers included the skewed items such as “offensive” or “obscene,” which have mainly negative connotations among others to measure provocation in advertising. We argue that each item used for measuring provocation should be general so as to be applicable to many types of provocative advertising and should not be skewed to a negative or positive side. Therefore, we checked the manipu-lation of the concept of provocation with responses to three general items: shocking, provocative and controversial.

 The second independent variable is the embedment of a social cause addressed in the ad. While companies have an unequivocal purpose of doing business, selling their products, their advertisements often address a social cause. In the current study, we embedded a social cause in the ad in the form of an advertising campaign theme. For example, we chose the advertising theme “Stop child soldiers!” for the provocative ads which address a social cause to be suitable to the advertising images we selected for the experiment. We will explain the selection process in more detail in the later part of this section.

 The dependent variables of the study were elaboration, ratio of negative thoughts to the total number of thoughts, activation of persuasion knowledge, attitude toward the ad, attitude toward the brand and purchase intention. The first two dependent variables were calculated from the number of cognitive responses to a thought-listing question after a subject was exposed to stimuli ads. Cognitive responses in this study were defined as any thought that passed through a person’ mind when he or she is exposed to the ads in the questionnaire (Cacioppo & Petty, 1981; Park, 2011). After asking subjects to list any thought that they had during the experiment, we also asked them, on the next page of the questionnaire, to categorize their own responses into three types: positive, negative, or neutral (DeBono & Harnish, 1988).

  More specifically, the first dependent variable, the level of elaboration, was computed by counting all cognitive responses a subject listed (e.g., Aaker & Sengupta, 2000; Miniard, Bhatla, & Rose, 1990; Mothersbaugh, Huhmann,& Franke, 2002). In other words, without considering categorization of the thoughts, a subject was considered to have higher elaboration on the ad if he or she listed more thoughts about the ad or the brand. The second dependent variable, the ratio of negative thoughts to the total number of thoughts, was computed by subtracting the number of positive thoughts from the sum of the total number of thoughts and the number of negative thoughts and divided by the total number of thoughts (Cacioppo & Petty, 1981). The third dependent variable, activation of persuasion knowledge, was measured with a subject’ agreement to a statement (When I saw the ads, I thought it was pretty obvious that the company was trying to make a sale.) (e.g., Campbell & Kirmani, 2000; Williams, Fitzsimons, & Block, 2004).

 The attitude toward the ad and attitude toward the brand were assessed with three seven-point items (Bad-good, unfavorable-favorable and unpleasant- pleasant). Purchase intention was assessed with three seven-point items (Very unlikely-very likely, definitely would not-definitely would and uncertain-certain) to the question of likelihood of purchasing the brand in the future. The moderating variable, the involvement with the issue of child welfare was measured with ten items (e.g., important-unimportant) of a semantic differential scale (Zaichkowsky, 1994).

 Selection of Stimulus Images

  Previous studies have often used actual ads to investigate the effect of provocative advertising (Dahl et al., 2003; Vezina & Paul, 1994). Although actual ads offer believability to the study, the possibility of many confound-ing variables weakens the validity of the study. These variables include the history of a brand, previous media coverage, store availability, personal experience with the brand or the ad, likeability or familiarity with an advertising model, etc. In this study, we created four types of ads for the experiment and put a fictitious name to the brand (Solarot) in the ads to minimize the effect of confounding variables.

  First, we decided children’ clothing as the product category for the experiment because the issue of child welfare was used in the stimuli ads. We expected that a universal value like child welfare would cause less unique responses among study participants. Furthermore, many clothing companies and nonprofit organizations for child welfare use provocative advertising images so the participants would be accustomed to this type of advertising (Pope et al., 2004; Szykman, Bloom, & Blazing, 2004). We, three authors, independently searched for four types of images of children for the experiment. Specifically, we looked for about ten images for each type: provocative images which address a social cause, provocative images which do not address a social cause, non-provocative images which address a social cause and non-provocative images which do not address a social cause. After collecting 40 images independently, we had a series of meetings to discuss criteria for each type of ads to best serve the purpose of the current experiment while minimizing the effect of possible confounding variables. We narrowed the number of images to 8 for each type resulting in a total of 32 images. Two graduate assistants who had the appropriate capability of editing digital images created 32 advertisements for the two rounds of pre-tests and they edited the stimuli ads throughout the research process.

  In the first pre-test, we asked subjects how provocative they considered each ad and whether the ad addressed a social cause. We used three items to measure advertising provocation and also three items to see whether the ad addressed a social cause (e.g., The company is trying to improve child welfare.). Based on the results, we kept the most proper ads for each type and created additional ads to keep the necessary number of ads for the second pre-test. At this stage, we made sure to have the same social cause among all the stimuli ads in the same category. For example, because we were not specific about the type of a social cause in the beginning of the selection process, the provocative images we selected for the first pre-test were about child poverty, child labor, child trafficking, child soldiers, etc. After the five images which addressed the same social cause were selected for each of the four categories, we administered the second pre-test to people who had not participated in the first pre-test. In the second pre-test, we added another question in which subjects were asked to write down possible ad campaign themes for the company in the ad.

  After reviewing the responses of the subjects and having a series of discussions, we selected three ads for each category and decided the ad campaign theme for each category: “Stop child soldiers!” for the provocative ads which addressed a social cause, “Make your child the next beauty queen!” for the provocative ads which did not address a social cause, “Stop child poverty!” for the non-provocative ads which addressed a social cause and “Let children be children!” for the non-provocative ads which did not address a social cause.

  In the main experiment, the questionnaire was composed of the introductory information about the company and advertising campaign, three ads, and questions. In a university classroom setting, one of the authors explained to student subjects that the purpose of the study was to investigate consumer responses to the ads of a European children’ clothing company. The researcher debriefed the real purpose of the study after the subjects filled out the questionnaire. The total number of subjects for the main experiment was 182. Three subjects who filled out the questionnaire inattentively and four more subjects who responded that they had seen the ads before were excluded from the final analysis. Therefore, a total of 175 subjects were used in the main analysis. The average age of the subjects was 21.0 and the percent of male subjects was 60.2.

  First, we checked the internal consistencies for all variables which had multiple items. The internal consistency of the ten items which measured a person’s involvement with the child welfare issue was .879 so we kept all ten items to create a summated variable. The Cronbach’s alpha of the three items to measure the provocative level of stimuli ads was .927 so it was satisfactory (Nunnally, 1978). The three items used to measure the attitude toward the ad and attitude toward the brand also showed satisfactory internal consistencies (.938 and .957 respectively). However, one item (Uncertain-certain) for measuring a person’s purchase intention made the internal consistency of the variable decrease considerably so this item was excluded from the summated variable and the alpha for the remaining two items was improved originally from .828 to .946.

  We compared the level of perceived provocation of the stimuli ads among the four conditions. As we manipulated, the provocation levels in the two provocative ad conditions (With a social cause, M = 6.1, SD = 1.0 and without a social cause, M = 5.6, SD = 1.6) were perceived to be higher than those in the two non-provocative ad conditions (With a social cause, M = 3.5, SD = 1.2 and without a social cause, M = 2.3, SD = 1.4). Because we randomly distributed four types of questionnaires, the involvement level with the child welfare issue was expected not to be significantly different among the four groups and it was not (F(3, 171) = 2.042, p = .110).

Results

  As the first hypothesis postulated, people had more elaboration on provocative advertising than on non-provocative advertising (F(1, 171) = 6.575, p < .05) and no statistical significance was found for the main effect of the social cause variable and the interaction effect. As stated in the method, we computed people’s elaboration level by summing all of their thoughts. The subjects who were exposed to the provocative ads which addressed a social cause listed an average of 5.0 (SD = 2.5, n = 47) thoughts, people with the provocative ads which did not address a social cause 4.8 (SD = 2.4, n = 48), people with the non-provocative ads which addressed a social cause 4.2 (SD = 1.9, n = 34) and people with the non-provocative ads which did not address a social cause 3.9 (SD = 1.6, n = 46). Therefore, hypothesis 1 was supported.

  Hypothesis 2 expected that the issue involvement would moderate the effect of the perceived provocation on people’s elaboration level on the ads. We first split the file depending on the embedment of a social cause in order to test the hypothesis in both cases. Whether or not there was a social cause in the ad, no statistical significance was found for this interaction effect. Therefore, hypothesis 2 was not supported. Further analysis found that in the social cause condition, people’s elaboration level on the ads was not significantly different whether or not they were involved with the social issue addressed in the ad. However, in the no-social cause condition, people had more elaboration on the ads when they were more involved with the issue (F(1, 90) = 4.327, p < .05).

  Hypothesis 3 postulated that there would be an interaction effect between the two independent variables on the activation of persuasion knowledge. Two-way ANOVA showed that there was an interaction effect between the provocation level in advertising and the embedment of a social cause in the ad (See Figure 1, F(1, 171) = 4.432, p < .05). A closer look indicated that this interaction effect was primarily caused by the large difference between those who were exposed to the provocative ads which addressed a social cause (M = 3.1, SD = 2.0) and those who were exposed to the provocative ads which did not address a social cause (M = 4.5, SD = 2.0). Therefore, hypothesis 3 was supported.

Figure 1. Activation of Persuasion Knowledge in the Four Conditions

  As can be seen in Figure 2, an interaction effect was found between the provocation level in advertising and the embedment of a social cause on the ratio of negative thoughts to the total number of thoughts (F(1, 171) = 6.781, p < .05). Therefore, hypothesis 4 was supported. In the provocative ad conditions, those who were exposed to the ads which did not address a social cause (M = 1.6, SD = .6) had a higher ratio of negative thoughts than those who were exposed to the ads which addressed a social cause (M = 1.3, SD = .7). However, in the non-provocative ad conditions, those who were exposed to the ads which addressed a social cause (M = 1.1, SD = .6) had a higher ratio of negative thoughts than those who were exposed to the ads which did not address a social cause (M = .8, SD = .7).

Furthermore, this two-way ANOVA revealed that those who were exposed to the provocative ads (M = 1.5, SD = .7) listed a higher ratio of negative thoughts than those who were exposed to the non-provocative ads (M = .9, SD = .7, F(1, 171) = 28.620, p < .001).

Figure 2. Ratio of Negative Thoughts in the Four Conditions

  We expected in hypotheses 5a to 5c that there would be an interaction effect between the two independent variables on three attitude measures. In the attitude toward the ad, an interaction effect was found (F(1, 171) = 10.698, p < .01) so that hypothesis 5a was supported (See Figure 3). In the attitude toward the provocative ads, people reacted more positively to the ads if the ads addressed a social cause (M = 3.3, SD = 1.8) as com-pared to those who were exposed to the ads which did not address a social cause (M = 2.3, SD = 1.6). However, in the attitude toward the non-provocative ads, people reacted more positively to the ads if the ads did not address a social cause (M = 4.4, SD = 1.5) than those who were exposed to the ads which addressed a social cause (M = 3.9, SD = 1.4). Along with the interaction effect, the effect of provocation in advertising was found to be statistically significant for the attitude toward the ad (F(1, 171) = 31.317, p < .001). Specifically, people in the provocative ad conditions showed a more negative attitude toward the ad (M = 2.8, SD = 1.8) than people in the non-provocative ad conditions (M = 4.2, SD = 1.5).

Figure 3. Attitude toward the Ad in the Four Conditions

  As expected in hypothesis 5b, an interaction effect was also found in the attitude toward the brand (See Figure 4, F(1, 171) = 6.533, p < .05). For the attitude toward the brand in the provocative ad conditions, people responded more positively to the brand if the brand was shown in the ads which addressed a social cause (M = 3.8, SD = 1.8 vs. M = 2.7, SD = 1.7 for the no-social cause condition). For the non-provocative ad conditions, people showed a more positive attitude toward the brand if the ads did not address a social cause (M = 4.6, SD = 1.4 vs. M = 4.4, SD = 1.3 for the social cause condition). Therefore, hypothesis 5b was supported. As we found in the analysis of the attitude toward the ad, the main effect of the provocation level in advertising was statistically significant for the attitude toward the brand (F(1, 171) = 27.928, p < .001; M = 3.2, SD = 1.8 for the provocative ads and M = 4.5, SD = 1.4 for the non-provocative ads). The embedment of a social cause also had a main effect on the attitude toward the brand (F(1, 171) = 4.005, p < .05). That is, people who were exposed to the ads which addressed a social cause (M = 4.0, SD = 1.6) showed a more positive attitude toward the brand than those who were exposed to the ads which did not address a social cause (M = 3.6, SD = 1.8).

Figure 4. Attitude toward the Brand in the Four Conditions 

   An interaction effect between the two independent variables was found on purchase intention as expected in hypothesis 5c (See Figure 5, F(1, 171) = 4.148, p < .05). In the provocative ad conditions, people had a higher purchase intention when the ad addressed a social cause (M = 3.2, SD = 2.0 vs. M = 1.9, SD = 1.2 for the no-social cause condition). Even in the non-provocative ad conditions, people had a higher purchase intention when the ad addressed a social cause (M = 3.4, SD = 1.6 vs. M = 3.0, SD = 1.5 for the no-social cause condition). Along with the interaction effect, the provocation level in advertising had a main effect on purchase intention (F(1, 171) = 7.083, p < .01). More specifically, people in the provocative ad conditions (M = 2.5, SD = 1.8) showed a lower purchase intention than those in the non-provocative ad conditions (M = 3.2, SD = 1.5). The embedment of a social cause also had a main effect on purchase intention (F(1, 171) = 13.429, p < .001). That is, people had a higher purchase intention when the ad addressed a social cause (M = 3.3, SD = 1.8 vs. M = 2.4, SD = 1.5 for the no-social cause condition).

Figure 5. Purchase Intention in the Four Conditions

Discussion

  Due to the increased advertising clutter in recent years, advertisers are more frequently relying on provocative advertising strategies to reach their audience. The current study was conducted to better understand why some provocative ads are more effective than others and what makes them more effective. The study confirmed that provocative ads made people think more than non-provocative ads. Although we did not directly measure people’ recall and recognition, we could infer that provocative ads would bring about higher recall and recognition of the ads and the brand because elaboration and recall have been proven to be highly correlated (Celsi & Olson, 1988; Huhmann, 2007; MacInnis et al., 1991).

  The result showed that people who were exposed to the provocative ads which did not address a social cause drew on their persuasion knowledge more than any other group. This suggests that if consumers could not read a social cause addressed in provocative ads, they are more likely to infer the commercial motive of an advertiser. Therefore, it is advisable for an advertiser to let consumers recognize, how subtle or overt the ad might be, that the company is trying to address a certain social issue if the company actually intends to address the issue. The company could use a variety of methods to achieve this goal including a press release, text copy given in the ad, etc. Based on the findings, we could infer that even though all forms of provocative advertising might incite people to draw on their persuasion knowledge, the activation of persuasion knowledge might be suppressed when a social cause is addressed in the ad.

  The interaction effect between provocation in advertising and the embedment of a social cause in the ad on the ratio of negative thoughts indicated that people had a higher ratio of negative thoughts if they perceived the ad as purely provocative. In other words, although provocative ads, in general, brought about more negative thoughts than non-provocative ads, this effect tended to be reduced by the embedment of a social cause. Therefore, it would be wise for an advertiser to instill a social cause in provocative ads to buffer the company against possible consumer criticisms.

  When people perceived an ad as provocative, their attitude toward the ad and the brand was more negative than that of people in the non-provocative ad conditions. This result is consistent with most previous research findings in the field. However, this negative attitude was considerably weakened when the ad addressed a social cause. The social cause in provocative ads keeps consumers from thinking these types of ads to be negative. Of course, there will be other aspects of provocative ads besides the embedment of a social cause which do not warrant a negative attitude toward the ad and the brand. Future research could examine what other factors deter consumers to react negatively to provocative ads.

  In the current study, we concurred with previous studies that provocative advertising generates a negative attitude toward the ad and the brand. The dilemma of companies, however, is that people remember provocative advertising better than non-provocative advertising as confirmed in the higher elaboration of provocative advertising (See also Bello et al., 1983; Dahl et al., 2003; Vezina & Paul, 1994). If consumers could not remember the ad and the brand in non-provocative ads, a positive attitude toward the ad and the brand would be a disputable advantage of using non-provocative ads. However, if a social cause is addressed in provocative ads, the provocative strategy becomes a clearly better option for the companies. The current study reveals that provocative advertising which addresses a social cause not only decreases the ratio of negative thoughts to the total number of thoughts, but also it brings about a higher purchase intention as much as non-provocative advertising.