Content-Type: text/html The Effectiveness of Comparative Advertising Among Koreans: Is It Effective to Increase the Intensity of Comparison over Time? THE EFFECTIVENESS OF COMPARATIVE ADVERTISING AMONG KOREANS: Is It Effective to Increase the Intensity of Comparison over Time? Jang-Sun Hwang Doctoral Student University of Tennessee, Knoxville Mariea Grubbs Hoy Associate Professor University of Tennessee, Knoxville Contact Person: Jang-Sun Hwang 426 Communications Bldg. University of Tennessee Knoxville, TN 37996 [log in to unmask] 865-946-5284 2000 Annual Convention, Phoenix, AZ The Association for Education in Journalism and Mass Communication THE EFFECTIVENESS OF COMPARATIVE ADVERTISING AMONG KOREANS: Is It Effective to Increase the Intensity of Comparison over Time? ABSTRACT Despite the popularity of comparative advertising (CA), few studies have explored how to develop this strategy outside the United States. This study reports the results of an experiment conducted in South Korea, a country where CA is rarely used. Two hundred Korean college students were exposed to fictitious advertisements in which the independent variable of comparison intensity (non-comparative/low/medium/high/increasing) and exposure sequence (first/second/third) were manipulated. The dependent variables of attitude toward the brand and purchase interest assessed advertising effectiveness. Results suggested that the effectiveness of comparative advertising in Korea could be improved by increasing the intensity of comparison gradually from indirect (low) to direct (high) over time. INTRODUCTION During the past quarter century, comparative advertising (CA) has been one of the most popular advertising message formats in the United States since the Federal Trade Commission's informal encouragement of explicit comparisons in 1972 (Barry 1993b). Numerous research efforts have been devoted to this area, and the majority of CA studies have explored the effectiveness of CA by contrasting CA with non-comparative advertising (NCA). Although the effectiveness of CA, according to the large body of extant empirical research, seems to be equivocal, most researchers concur that CA is more effective than NCA under some conditions. The CA research seems to be exhausted in terms of finding critical variables that influence consumers' responses. Several recent studies summarized and conceptualized the outcomes of CA studies during the last 25 years (Barry 1993a; Barry 1993b; Grewal et. al. 1997). Previous CA studies can fall under the combinations of three kinds of variables as shown in . Dependent variables fall under the three broad categories of consumer response: cognitive, affective and conative (Lavidge and Steiner 1961). Independent variables can be categorized as either external or internal. External dimensions are beyond the control of the CA execution itself. For example, "product types" and "market position (of the sponsored brand)" are aspects that the CA message cannot modify. Moreover, these variables are inevitably shown implicitly or explicitly in the ad. Internal dimensions can be modified by CA executions. "Source credibility" has been one of the most extensively researched internal aspects. Nonetheless, some issues still have received limited attention. First, since previous CA research was conducted in the U.S. among American consumers, few studies argue that their findings are applicable outside the United States. Within other cultures, both external and internal independent variables can be greatly effected due to different circumstances such as different consumers under different legal status of CA. A recent study focusing on cross-cultural issues of CA argued that CA is not a universally accepted advertising format due to consumers' < TABLE 1 > The Major Variables and their Literature Variables Representative Studies Dependent Variables Cognitive ù Attention ù Awareness (Recall) ù Information processing ù Believability Wilkie and Farris (1975), Barry and Trambly (1975), Prasad (1976), Pride et. al. (1977), Jain and Hackleman (1978), Grossbart et. al. (1986), Pechmann and Stewart (1990b) Affective ù Attitude toward the ad ù Attitude toward the product Belch (1981), Swinyard (1981), Etgar and Goodwin (1982), Shimp and Dyer (1978), Droge (1989), Muehling (1987) Conative ù Intention to purchase Shimp and Dyer (1978), Iyer (1988), Droge (1989) Independent Variables External Product Type ù Functional vs. Psychological ù Involvement (High vs. Low) Etgar and Goodwin (1977), McDougall (1978), Ash and Wee (1983), Droge (1989), Muehling et. al (1990) Market Position ù Low-share vs. High-share ù New vs. Existing Iyer (1988), Pechmann and Stewart (1991a), Droge and Darmon (1987), Gnepa (1993), Auken and Adams (1998) Cross-culture ù CA Usage (Dominant vs. Rare) Donthu (1998) Internal Source Credibility ù High vs. Low Gotlieb and Sarel (1991), Belch (1981), Boddewyn (1978), Kelly (1972) Message Type ù One sided vs. Two sided Swinyard (1981), Etgar and Goodwin (1982) Types of Comparison ù Superiority vs. Parity Buchanan and Smithies (1989), Barry (1993b) reluctance and the illegal status of CA in some countries (Donthu 1998). According to Donthu's research, in some countries like Hong Kong, Belgium, and South Korea, CA is illegal. Moreover, even though it is legally allowed, CA format is not frequently used in many countries outside the U.S. It is questionable whether CA is as effective in other countries as much as it is in the U.S. Second, previous CA studies lack a long-term perspective. Most CA studies have been pre-post experimental research designs (e.g. Pride et al. 1977; McDougall 1978; Shimp and Dyer 1978). That is to say, CA researchers examined consumers' responses, which are dependent variables, at one-time (one-shot study) rather than conducting multi-shot study. Therefore, these studies cannot address the managerial implications regarding long-term management of a CA campaign. Lamb, Pride and Pletcher (1978) proposed a taxonomy of comparative advertisements. They argued that the characteristics of comparative advertisements are determined by the intensity of comparisons for the most part. In their taxonomy, the intensity of comparison has three levels - "low," "moderate," and "high." In the long-term perspective, it is important to consider which one of the three levels of CA would be effective for inaugurating a campaign and which one would be useful for the last period of an advertising campaign. Examining the cultural issues in light of Lamb, Pride and Pletcher's (1978) taxonomy may prove beneficial. The consideration of the intensity of comparison of CA can reduce barriers to executing CA in some countries where the CA format has traditionally not been popular. According to a recent cross-cultural CA study (Donthu 1998), people of countries that have rarely been exposed to CA formats are more likely to have negative attitudes toward the advertised brands as well as the comparative ads themselves. The main consideration involves reducing consumers' negative attitudes toward CA formats in those countries. Some countries that had banned CA have recently permitted its use. South Korea is one of those countries. Traditionally, Korean consumers have been often exposed to indirect CA (claiming the superiority or parity of the sponsored brand without referring any competing brand's name), but rarely to direct CA (explicit reference of any competing brand's name). The purpose of this study was to investigate the effectiveness of presenting comparative advertisements that vary in intensity of comparison over time to Koreans as determined by their affective and behavioral intention responses. THEORETICAL FRAMEWORK AND RESEARCH DESIGN Repetition and Strength of Advertising Claims Although several studies have confirmed the positive effect of repetition and strength on cognitive evaluations (Crowder 1976; Singh et al. 1994; Pechmann and Stewart 1988; Singh, Rothchild, and Churchill 1988), there is no consensus as to whether the effect is still valid for the affective evaluation. Some studies reported that attitudes toward messages could be more preferable as the number of repetition is increased (i.e., Schumann, Petty, and Clemons 1990; Belch 1982; Cacioppo and Petty 1979; Zajonc 1968). Others focusing on "wear-out" phenomenon insist that as a new stimulus is repeated, people are likely to have negative attitudes toward it due to the monotonous mood and tedium (Berlyne 1978; Mitchell and Olson 1977; Ginter 1974; Winter 1973). Nevertheless, most research supports the idea that the repetition of message could decrease the uncertainty and conflict formulated by message. Generally, strong arguments seem to be more effective in persuasion than weak arguments. In the process of information gathering, strong arguments are expected to yield preferable cognitive and affective responses, while weak arguments are to result in counter-argument or negative responses (Petty and Cacioppo, 1981; Petty, Cacioppo, and Schumann 1983). A study examining the intensity of arguments (Petty, Cacioppo, and Heesacker 1981) suggested that strong arguments were perceived as more logical, reinforced and fair than weak arguments. Furthermore, CA research has rarely addressed this concern. Although most CA studies have talked about the effectiveness of the CA format (strong argument) compared to the non-comparative format (weak argument), few studies have concentrated on the variation of intensity of comparison. A few studies (Pechmann and Stewart 1991; Pechmann and Ratneshwar 1991) explored the comparison between direct CA - strong argument - and indirect CA - weak argument, and they agreed upon the fact that direct CA would be more effective for cognitive and affective responses than indirect CA. The reason why only a few studies focused on the variety of CA in terms of the intensity of comparison is mainly due to the lack of clear definition of CA. In fact, while most of the early researchers in this field included direct CA in the category of CA format (e.g. Levine 1976; Prasad 1976; Pride et al. 1977), some researchers considered indirect CA as a type of CA and compared it to NCA (e.g. Shimp 1975; McDougall 1977; Wilson 1978; Ash and Wee 1983). This study takes a somewhat broad definition of CA by including indirect as well as direct CA formats. Lamb and his colleagues (1978) provided an advanced and broader perspective toward comparative advertising (CA) by proposing a taxonomy of CA. Their taxonomy has two factors that determine the characteristic of CA: "Direction (Association or Differentiation)" and "Intensity of Comparison (Low-Medium-High)." The direction differentiates two types of CA claims: "stressing similarity message" and "stressing differentiation message." This study focuses on the intensity of comparison rather than the direction of message. A modification of Lamb, Pride and Fletcher's (1978) taxonomy is shown in < TABLE 2 >. Differentiation and association in terms of the direction of message can apply to these three categories. The conceptualization of each category stems from the overlapped concepts of differentiation and association messages. < TABLE 2 > Taxonomy for Comparative Ad by the Intensity of Comparison. LOW The advertisement does not identify the competing brand(s) by name, but casually refers to it in other ways such as "the leading brand." MEDIUM The name of the competing brand(s) is identified but is mentioned infrequently. The comparison does not occur on a point-by-point basis. HIGH The competing brand name(s) is identified and is mentioned frequently in a point-by-point comparison. Note) This taxonomy adopted and modified Lamb, Pride, and Fletcher's (1978) taxonomy. Carry-Over Effects of Advertising Message With the long-term perspective, it has been a critical issue among advertising practitioners how much they could draw from the advantages or disadvantages from previous ad campaigns to the current one. Advertising effects have been generally reported to accumulate over time rather than appearing and disappearing in a short time (Belch 1982). If this is true, the following critical questions should be addressed: "How long are the ad effects maintained continuously?" or "How much do the ad effects decrease over time?" However, this carry-over effect can vary by the type of brand (new vs. existing), the consumer's information processing ability, the degree of memory reinforcing, the product category, and the time of research execution. Krugman (1972) empirically examined the carry-over effect of advertising, and concluded that consumers require at least three times of exposures to an ad in order to influence their intention to buy the advertised product. According to his explanation, when first exposed to an ad, a consumer responds "What is it?" At the second exposure, (s)he thinks "What does that product mean to me?" The consumer may decide whether to purchase at the third exposure. Although the purchase decision is depending upon all other marketing mix factors such as price, quality, and package (Clarke 1976; Palda 1964; Sawyer and Ward 1977), it is generally accepted that there is a kind of carry-over effect, and most media planners employ this concept, "effective reach as 3," in their media scheduling strategy. Therefore, exposure frequency is an appropriate independent variable for examining the impact of comparative advertising. South Korea and Comparative Advertising Usage As Donthu (1998) examined, South Korea as well as most Asian countries has not legally allowed comparative advertising (CA), especially direct CA for a long time. Regardless of its legal status, the advertising communities of those countries have implicit understanding that CA will not be used. Advertisers fear the counter-effect against their brand image if they use CA format, which is traditionally perceived as taboo. Traditionally, Buddhism and Confucianism have become the philosophical foundations for the Korean value system. Within the context of Buddhism, the possession of material things and other secular desires have been considered trivial (Kim 1996). Since the fifteenth century, Confucianism has also been thoroughly institutionalized and systematically diffused into the Korean society (Yum 1987). This philosophy defines proper human relationships as the basis for a just society, and also emphasizes the value of cooperation rather than competition (Kim 1996). As a result, the virtue of benevolence and unobtrusive manner permeates Korean culture. In fact, Confucianism has influenced Korean religious life for a long time, and has had an overwhelming effect on the social and political aspects of the culture. Therefore, the deeply ingrained Confucian mode of manners and social relations is still a major factor in the way Koreans think and act (Korean Overseas Information Service pp.137-140). One of the disciplines this philosophy emphasizes is the goodness of unobtrusiveness; therefore, Koreans have regarded it ill-mannered behavior to argue one's comparative merits against others (Osgood 1951). This chronic ideal has made CA be interpreted as criticizing and attacking other competitors through advertisement - even with correct information. As such, CA is considered "unethical" in the Korean value system (Kim 1996). However, this country has rapidly been westernized especially during the last three decades. Therefore, the government has deregulated the CA format recently. During the last decade, Korean consumers have become increasingly exposed to some indirect advertisements, although most were regulated by government based on the "unfairness" rule. Finally, since July 1, 1999, a new advertising act has been inacted. Based on this new policy, advertisers can compare with other competitors with specific information claims, only if the information is objective and correct (Korean Broadcasting and Advertising Corporation 1999). The most important change due to this policy is that Korean advertisers can name the competitors. In fact, during the last several decades, the Korean Fairness Trade Committee - a FTC of Korea - strictly regulated some comparative advertisements, even though most were indirect CA. Recently, some advertisers tried to execute indirect CA format and gained high effectiveness. For example, Asiana Airlines, the second largest airline, challenged against their competitor, No.1 Korean Air, with a CA campaign. Although it was an indirect CA format without referring to the name of competitor, most consumers would know to whom Asiana Airlines referred. Because there have been only two airlines in South Korea, their indirect CA was not much different from direct CA format. Their headlines were "Could you take a new plane, or an old one?" They stressed on their comparative advantage over their competitor; in fact, their planes were less than 10 year old ones, while the leading brand's planes were more than 20 years old. Although their claim was not deceptive and based on the fact, the competitor accused Asiana Airlines, and the argument was seriously continued for a while. Although there is a governmental moderator, the Fair Trade Committee - a FTC of Korea - mediating this kind of problem, the effectiveness of its authority has not clearly solved problems. Moreover, once being involved in the legal argument, the brand image of sponsored product as well as the competitor's one was usually damaged. Although the new advertising policy allowed advertisers to execute direct CA, advertisers are still reluctant to use direct CA due to the social atmosphere, which has been traditionally built by Buddhism and Confucianism as previously described. Therefore, it is a critical issue how to employ CA format without alienating the target consumers or damaging one's brand image. Under some situations, it has been reported that indirect CA is more effective than non-CA (NCA) for cognitive responses in Korea (Lyi 1988; Kang 1989). However, it is unknown whether the message of strong intensity of comparison, direct CA, is effective for affective or behavioral intention dimensions in Korea. Based on the previously discussed concepts, "the intensity of comparison" and "three exposures for carry-over effect," the current study focuses on the managerial effectiveness of CA message variations in terms of the intensity of comparison over time. Specifically, this study examines the effectiveness of the message presentation from the lowest level of comparison intensity to the highest one. The former is familiar to Korean consumers, but the latter is not. Therefore, this study explored if Korean consumers can adapt themselves to this relatively new format of advertisement. Specifically, affective as well as behavioral intention dimensions of Korean consumers' evaluations of the advertised product were explored. The following hypotheses are expected to explore the long-term effect of CA with the variation of the intensity of comparisons: H1: There will be some significant differences in the attitudes toward (ABR) and purchase interests (PI) on the sponsored product based on the intensity of comparison. H1a: Comparative ad will be more effective than non-comparative ad. H1b: As intensity of comparison increases, the ad will be more effective regardless of exposure. H2: Ad effectiveness will increase over time. H3: There will be an interaction between exposure and intensity of comparison: As intensity of comparison increases from low to high, ad effectiveness increases (ABR , PI) METHODOLOGY The independent variables were manipulated in a three shot pre-post experiment with level of comparison intensity (non-comparative/low/medium/high/increasing) and exposure (first/second/ third) - repeated measure. Two hundred Korean college students attending a private English class during winter vacation served as the convenience sample. The procedure was conducted three times over nine days with 3-day intervals between each data collection. Special concern was paid to the maintenance of samples. With the consent of lecturer, all students in the class voluntarily participated in the first test, and 180 provided complete data for all three exposures and thus provided the usable sample. The balance among the five experimental groups (See FIGURE 1) was more or less achieved. In fact, completing three-time experiment, the number of valid participants within each treatment group was ranged from 34 to 40. Since the class was held during vacation and all students voluntarily took the class, the concern of sample maintenance was the most critical issue in this study. To encourage attendance and participation, small accessories were given to all participants as compensation after each data collection procedure. Prior to data collection for each of the three exposures, confidentiality was emphasized. In order to identify the subjects' questionnaires, each participant was assigned an individual identification number. Therefore, researchers could match their three questionnaires and know which of the five treatments they had experienced. To encourage subjects to report their identification number for each of the two subsequent exposures, demographic questions were omitted. However, some descriptions of the sample in terms of demographic aspects is possible. In general, the subjects were 18-24 and included both men and women. The gender composition was roughly even. During the first data collection procedure, subjects were randomly assigned to one of the five experimental groups representing the five levels of comparison intensity. < Figure 1 > shows the experiment process over the three exposures. The non-comparative advertisement served as the control. Fictitious advertisements representing the manipulations of comparison intensity were presented to subjects. The test advertisements expressed the levels of comparison intensity in both visual and textual components. Each subject saw only one ad on each separate day of data collection. Intensity of comparison was constant across the 3 ads (except for the "increasing intensity" treatment). However, due to the forced exposure of the stimulus and resulting concern that the students might get bored which would influence their attitude scores, slight differences in copy were presented (See Appendix). In fact, it has been regarded to be the threat to internal validity to do repeated measures with maintaining subjects, for the problem of maturation can happen within the respondents over time (Kerlinger 2000 p.470). In order to minimize this threat, it was necessary to have some cosmetic variations of advertisements over time within the same groups. Subjects were allowed to see each ad for 2 minutes. After receiving a self-administered questionnaire that assessed affective and behavioral measures, subjects were not allowed to return to the ad. < FIGURE 1 > The Process of Experiment Groups 1st Exposure 2nd Exposure 3rd Exposure CONTROL NCA A ( NCA B ( NCA C LOW LOW A ( LOW B ( LOW C MEDIUM MED A ( MED B ( MED C HIGH HIGH A ( HIGH B ( HIGH C INCREASING LOW A ( MED B ( HIGH C Note) LOW A Means the "first (A) comparative ad with low intensity of comparison Development of Ad Stimuli for Manipulation of Comparison Intensity The product used in the experimental ads was a notebook computer. First, personal computers were taken as the experimental product category, for it is generally regarded that the participants of this study, the college students, frequently use this product, regardless of gender. Prior to this main experiment, a pre-test was conducted to determine the type of computer and the competing brand with 60 Korean college students. The results showed that approximately half of them (48.3%) wanted to buy notebook computers followed by Pentium desktop computers (38.3%). (Regarding the competing brands, five major computer manufacturers were referred to and the leading brand of notebook computers, IBM, was ranked No.1 as the most favorable computer among them (75%). Therefore, IBM ThinkPad Notebook computer was selected as the competing brand. Advertisements for a fictitious brand of computer and its features were created. All features were made with thorough consideration to compete with IBM ThinkPad. For instance, the brand name was made in English using three characters (ESN: Easy Speed Notebook) in order to be similar to IBM. Based on the taxonomy proposed in
, each ad was manipulated with the variation of intensity of comparison. The ad messages presented took the differentiation type rather than the association one. Since differentiation is much more widely used form than association (Lamb, Pride, and Fletcher 1978), all ads used in this study employed differentiation formats. Non-comparative ads were shown not to claim any superiority (See Appendix B). The operational definitions of the three levels of comparison intensity and the control (non-comparative) are presented in < TABLE 3 >. < TABLE 3 > Operational Definitions for Levels of Comparison Intensity Level Descriptions Low Intensity CA Claiming superiority to any anonymous competing brands. Med Intensity CA Identifying the competing brand without detailed comparisons. High Intensity CA Identifying the competing brand with comparison in detail. Non-CA (Control) No claims for superiority or comparison Dependent Variables The questionnaire used for all three exposures was same. Three questions were designed to measure affective dimension, while other three questions measured the behavioral intention dimension. In order to verify the reliability of the test ads, last question was added and asked, "The ad is credible." Because the ad was fictitious, it was necessary to verify the subjects perceived the ad as credible. All seven questions used a 7-point bipolar Likert scale (1 - strongly disagree, 7 - strongly agree). The scales used for the dependent variables are summarized as follows. (1) Attitudes toward the sponsored brand (ABR) are composed of three sub measures which translate as follows: "Preference - I prefer the product of the ad," "Liking - I like the product of the ad, " and "Expectation - I expect the product of the ad is better than other brands." These three items were calibrated to measure the overall attitude toward the product. The mean score of these three measures was used for the overall attitudes toward the product. Cronbach's alpha coefficient values supported this decision. For each exposure, the Cronbach's alpha values were all larger than .70 (.76, .83, .89), the recommendable threshold to consider that all sub-categorical variables measure the same main variable (Nunnally 1978). (2) Purchase intentions of the sponsored product (PI) are also composed of three sub-categorical measures which translate as follows: "Likelihood of purchase - I am likely to purchase the advertised product in the future," "Expectation - I expect, the function of product is beneficial to me," and "Expected satisfaction - If purchased, the product can satisfy me with its merits." Cronbach's alpha coefficient value of each exposure was ranged from .71 to .92, which verifies these three items measure the same variable, the overall purchase intentions. ANALYSIS AND RESULTS Each hypothesis was analyzed by comparing means with the GLM (General Linear Model) framework procedure, which makes allowance for unequal cells in the ANOVA procedure (SAS Institute, pp.893-895). The following tables show the mean scores of each measurement by different treatment over time. Based on the outcomes shown in these tables, each hypothesis was examined. < TABLE 4 > Mean Scores of Attitudes toward the Brand and Purchase Interests Attitude Toward The Brand Means* Comparison Intensity Exposure Mean (H1) 1st 2nd 3rd CONTROL** (n=34) 3.69 3.88 3.99 3.86 LOW (n=34) 3.44 4.12 4.00 3.85 MEDIUM (n=37) 3.81 4.24 4.30 4.12 HIGH (n=35) 4.26 3.94 3.66 3.96 INCREASING (n=40) 3.72 4.23 5.43 4.46 Mean (H2) 3.78 4.08 4.31 Purchase Interest Means* Exposure Comparison Intensity Mean (H1) 1st 2nd 3rd CONTROL ** (n=34) 3.62 3.35 3.18 3.38 LOW (n=34) 3.35 3.75 4.16 3.76 MEDIUM (n=37) 3.42 4.26 4.21 3.96 HIGH (n=35) 4.28 3.96 3.62 3.95 INCREASING (n=40) 3.73 4.08 5.40 4.40 Mean (H2) 3.68 3.90 4.15 * Assessed on a 7-point scale where a higher means indicates a greater purchase interest. ** The Control Group saw a non-comparative advertisement. The Effectiveness of CA with the Intensity of Comparison (H1) First, the attitudes toward the sponsored brand (ABR) showed significant differences among five different treatments. Regardless of the exposure time, as shown in < Table 4 >, the "Increasing" group participants showed the most preferable attitudes toward the product (mean = 4.46) followed by "Medium (4.12)," "High (3.96)," "Control (3.86)," and "Low (3.85)" groups. Specifically, a post hoc test (Tukey) showed that the overall differences are resulted from the significant differences between "Increasing" - "Control" and "Increasing" - "Low" with significance level of .05. Similarly, "purchase interests (PI)" also showed significant differences among the five groups due to the large PI mean score of "increasing" group (4.40). As shown in < Table 5 > previously, the ranks of others are also almost same as ABR case: "Medium (3.96)" - "High (3.95)" - "Low (3.76)" - "Control (3.38)." Specifically, the "Increasing" treatment showed more significant effects on PI than "Control" and "Low" treatments. Overall, hypothesis 1 was moderately supported. Although all five treatments did not show significantly different effects on ABR and PI, the overall comparison among them verified significant differences. The "Increasing" treatment was the most effective one. < TABLE 5 > Comparison Intensity among Five Different Treatments ANOVA Results Attitudes Toward the Brand (ABR) Source df Sum of Squares Mean Square F Value Prob. > F Diff. ( p < .05) Model 4 9.607 2.40 3.04 .0187 Increasing - Control Increasing - Low Error 175 138.30 .79 Purchase Interests (PI) Source df Sum of Squares Mean Square F Value Prob. > F Diff. ( p < .05) Model 4 20.134 5.03 5.97 .0002 Increasing - Control Increasing - Low Error 175 147.647 .84 Additionally, analysis was conducted to compare the five experimental groups in each exposure time. Regarding attitudes toward the brand, for the first exposure, there was a significant difference for ABR (F=2.94 / p=.02) among five groups, specifically due to the significant difference between "High (mean=4.28)" and "Low (mean=3.35)." Although having no differences in the second exposure, significant differences was found in the third exposure (F=11.51 / p=.001) due to the large score of "Increasing" treatment. In fact, this group showed significantly higher effectiveness than the other four groups. The analysis of "purchase interests" in each exposure also showed similar results. All three exposures produced high F-values, thus satisfying the significance level of .05. Specifically, the result of the final exposure was almost same as the ABR case. In fact, the large score of "Increasing" group resulted in the significant difference among the different treatments. The Carryover Effects of CA (H2) Since the "exposure" of three shots was not a categorical variable, GLM (General Linear Model) allowed examination the carry-over effect. Specifically, the repeated measure function was used in the analysis. Therefore, the interaction between the three-time exposure effect and the variation of intensity of comparison was also examined in the same procedure of GLM. According to the result of comparison among the three exposures, both attitude toward the brand (ABR) and purchase interest (PI) showed significant differences over time < TABLE 6 >. That is to say, based on the increasing trend in < FIGURE 2 >, participants were likely to have more positive attitudes and intentions over time. The differences among the three groups (1-2, 2-3, and 1-3) were all statistically significant (p < .0001). Thus, hypothesis 2 was supported. < TABLE 6 > Comparison among Three Exposures and Its Interaction with Different Treatments Attitudes Toward the Brand (ABR) Source df Sum of Squares Mean Square F Value Prob. > F Exposure (H2) 2 21.877 10.94 16.40 < .0001 Interaction* (H3) 8 58.611 7.326 10.98 < .0001 Error (Exposure) 350 233.443 .67 Purchase Interests (PI) Source df Sum of Squares Mean Square F Value Prob. > F Exposure (H2) 2 16.90 8.45 10.56 < .0001 Interaction* (H3) 8 80.14 10.02 12.52 < .0001 Error (Exposure) 350 280.04 .80 * Interaction = Exposure x Treatment (Group) < FIGURE 2 > Changes of ABR and PI Over time The Interaction of Carryover Effect with the Intensity of Comparison (H3) Hypothesis 3 examined the interaction relationship between the "exposure" over three times and the intensity of comparison with five different treatments. As shown in < TABLE 6 >, the result showed that there was a significant interaction between the two variables for both attitude toward the brand (ABR) and purchase interest (PI). The significant interactions in ABR resulted from the score of "low intensity" group's first exposure score (3.44) and some the higher scores of "medium intensity (2nd exposure - 4.24 and 3rd exposure - 4.30)," "high intensity (1st exposure - 4.26)," and "increasing (2nd exposure - 4.08 and 3rd exposure - 5.43)." Especially, the mean of "increasing" group's final exposure (5.43) showed significant differences with all other interactions according to Tukey test results. The Tukey result in purchase interest also showed similar trend. The highest mean score among all fifteen interactions (3 exposures x 5 treatments) was the final exposure of "increasing" group (5.40), and this showed significant differences with all other interactions. Some other comparisons between interactions also contributed to this result. Those include interactions showing lower mean score of PI such as "control - 2nd (3.35) and 3rd (3.18)" "low - 1st (3.35)," and "medium - 1st (3.42)" as well as some interactions having higher scores of PI - "medium - 2nd (4.26) and 3rd (4.21)," "high - 1st (4.28)," "increasing - 2nd (4.08) and 3rd (5.40)." Therefore, hypothesis 3 was firmly supported. Analyzing by groups, as shown in < FIGURE 3 > and
, "control" subjects showed a decrease in their purchase interest scores, while their ABR scores increased. This contradictory result of control group can be congruent with the body of repetition effect literatures, which are also controversial on the effectiveness of message repetition as previously noted. The most interesting contrast resulted from the trend of "high" intensity group, in which participants showed the significant decreasing affective and behavioral intention scores over time, while the other groups receiving different comparison intensives - "low," "medium," and "increasing" - showed moderate or sharp increasing trends. Especially, "high" and "increasing" groups showed completely opposite trend over time in both dependent measures (ABR and PI). These two groups' contrasting trend contributed to the significant interaction between the "exposure" and "the intensity of comparison." Overall, the repetition of "high" intensity of comparison message caused negative effect, which is not congruent with the American literature, while the "increasing" treatment was proved to be effective toward the sample of this study: Korean college students. < FIGURE 3 > Changes of Attitude toward the Brand Score in each group over time < FIGURE 4 > Changes of Purchase Interest scores in each group over time Additionally, these overall findings are also reflected in the credibility test for each advertisement. The experimental group evaluated the final ad ("high" intensity) as the most credible when compared to other groups' evaluations on final ads. The only exception was the comparison of the "increasing intensity" group and the "high intensity" message group in the first exposure. In fact, "high" intensity group evaluated the first ad more credible than the "increasing" intensity group did on the first exposure ("low" intensity). This implies that CA of high intensity of comparison can be evaluated more credible than CA of low intensity of comparison. DISCUSSIONS AND MANAGERIAL IMPLICATIONS This study focused on a neglected area of CA study: varying intensity of comparison over time and cross-cultural reaction to comparative advertising. Based on the CA literature, this study assumed that CA is more effective than non-CA under some circumstances, and aimed to examine the effectiveness of CA by varying the intensity of comparison over time in terms of affective and conative responses. This study can be evaluated to a start in order to overcome the frequent limitations of previous CA studies with three-shot experiment. The focus was on an experiment group, which was exposed "low" intensity CA first followed by "mid" and "high" intensity CA. The current study mainly analyzed the changes of the "increasing" intensity group over time and compared it with other groups, which were exposed to the CA messages with maintaining the same intensity of comparison. In both affective and conative dimensions, the "increasing" group showed the most preferable attitude toward and the highest purchase interest in the sponsored product. Briefly, the CA strategy with gradual increases of the intensity of comparison over time was found to be effective compared to the strategy of maintaining the intensity to Korean consumers, especially the strategy of direct CA repetition ("high" intensity group). This study can yield some helpful implications for advertising practitioners who manage international advertising campaigns from a long-term perspective. They can benefit from the consideration of the management of the intensity of comparisons across different countries, which have different cultural and legal situations regarding CA. Specifically, international advertising planners need to execute relatively indirect comparative advertisements in the launching stage and to increase the intensity of comparison up to direct comparative advertisements when aiming at South Korean consumers. This study can provide a possible example of an effective strategy of advertising campaign with CA formats in the long run for consumers who have been rarely exposed to direct comparative advertising. It is needed to understand and adopt the underlying cultural values and situational factors such as government policy for most international advertising planners. Although aiming at some previously uninvestigated of CA research, this study has inevitable limitations, especially validity concerns. First, the representativeness of sample would be problematic. The college students can hardly represent all whole Korean consumers. Moreover, the sampling procedure was conducted as convenience sampling, which was gathered in a class. This may distort the truth of the population. Second, the management of each test was cumbersome work, and it could evoke some minor problems due to the respondent biases. Since only two researchers handled two hundred participants in each test, it could hardly take the benefit of experiment: "control." Further studies need to employ more researchers to supervise the large size of participants. Third, it took only one product category, which is a high-involvement product. The involvement has been reported a critical variable in CA effect research; however, this study overlooked this. Finally, the general limitation of experimental design can harm the external validity. One suggestion can be helpful for any further study to overcome this limitation. In-theater experiment can offer the "realistic" situation without Hawthorne effect, which resulted from the participants' recognition of being observed. REFERENCES Ash, Stephen B, and Chow-Hou Wee, "Comparative Advertising: A Review with Implications for Future Research," Advances in Consumer Research, Vol.10, pp.370~376. Auken, Stuart Van and Arthur J. Adams, "Attribute Upgrading through Across-Class Within-Category Comparison Advertising," Journal of Advertising Research, Mar./Apr. 1998, pp.6-16. Barry, Thomas E., "Comparative Advertising: What have we learned in two decades?" Journal of Advertising Research, Mar/Apr 1993, pp.19~29 (a). , "Twenty years of comparative advertising in the United States," International Journal of Advertising, Vol.12, 1993, pp.325~350 (b). , and Roger L. Tremblay "Comparative Advertising: Perspectives and Issues," Journal of Advertising, Vol.4, 1975, pp.15~20. Belch, George E., "An examination of comparative and noncomparative television commercials: The effects of claim variation and repetition on cognitive response and message acceptance," Journal of Marketing Research, Vol.18, Aug. 1981, pp.333-~349. , "The Effect of Television Commercial Repetition on Cognitive Response and Message Acceptance," Journal of Consumer Research, Vol. 9, June 1982, pp.56-65. Berlyne, D. E., "Novelty, Complexity, and Hedonic Value," Perception and Psychophysics, Vol. 8, 1978, pp.279-286. Boddewyn, J.J., and Katherin Marton, Comparison Advertising: A Worldwide Study, NY: Hasting House, 1978. Buchanan, Bruce and R. H. Smithies, "Substantiating a Parity Position," Journal of Advertising Research, Vol. 29, No. 5, 1989, pp.9-20. Cacioppo, John T. and Richard Petty, "Effects of Message Repetition and Position on Cognitive Response, Recall and Persuasion," Journal of Personality and Social Psychology, Vol. 37, Jan. 1979, pp.97-109. Clarke, Darral G., "Economic Measurement of the Duration of Advertising Effects on Sales," Journal of Marketing Research, Nov. 1976, pp.345-357. Crowder, R. G., Principles of Learning and Memory, Hillsdale, NJ: Lawrence Erlbaum Associates, 1976. Donthu Naveen, "A Cross-Country Investigation of Recall of and Attitude toward Comparative Advertising," Journal of Advertising, Vol. 27, No.2, 1998, pp.111-122. Droge, Cornelia, "Shaping the route to attitude change: central versus peripheral processing through comparative versus noncomparative advertising," Journal of Marketing Research, Vol.26, May. 1989, pp.193~204. , and Rene Y. Darmon, "Associative Positioning Strategies Through Comparative Advertising: Attribute Versus Overall Similarity Approaches," Journal of Marketing Research, Vol. 24, Nov. 1987, pp.377-388. Etgar, Michael and Stephen M. Goodwin, "Comparaison Advertising: Issues and Problems," Advances in Consumer Research, Vol.5, 1977, pp.63~71. , "One-Sided versus Two-Sided Comparative Message Appeals for New Brand Introductions," Journal of Consumer Research, Vol. 8, Mar. 1982, pp.460-465. Ginter, James L., "An Experimental Investigation of Attitude Change and Choice of a New Brand," Journal of Marketing Research, Vol. 11, Feb. 1974, pp.30-37. Gnepa, Tahi J., "Observations: Comparative Advertising in Magazines: Nature, Frequency, and a test of the Underdog hypothesis," Journal of Advertising Research, Sep./Oct. 1993, pp.70-75. Gotlieb, Jerry B. and Dan Sarel, "The Influence of Type of Advertisement, Price, and Source Credibility on Perceived Quality," Journal of the Academy and Marketing Science, Vol.20, No.3, 1992, pp.253-260. Grewal, Ddhruv, Sukumar Kavanoor, Edward F. Fern, Carolyn Costely, and James Barnes, "Comparative Versus Noncomparative Advertising: A Meta-Analysis," Journal of Marketing, Vol.61, Oct. 1997, pp.1-15. Grossbart, Sanford, Darrel Muehling, and Norman Kangun, "Verbal and Visual References to Competition in Comparative Advertising," Journal of Advertising, Vol. 15, No. 1, 1986, pp.10-23. Iyer, Easwar S., "The Influence of Verbal Content and Relative Newness on the Effectiveness of Comparative Advertising," Journal of Advertising, Vol. 17, No.3, 1988, pp.15-21. Jain, Subhash C. and Edwin C. Hackleman, "How Effective is Comparative Advertising for Stimulating Brand Recall?" Journal of Advertising, Vol. 7, No.3, 1978, pp.20-25. Kang, Il-Mo, A Study of the Accommodation of Comparative Advertising in Korea, Chung-Ang University, Seoul: Korea, 1989. Kelly, Harold H., "Attributions in Social Interaction," in Attribution: Perceiving the Causes of Behavior, Jones, E.E., e. Kanouse, H. Kelly, R. Nisbett, S. Valins, and B. Weiner, eds. Norristown, NJ: General Learning Press, 1972. Kerlinger, Fred N. and Howard B. Lee, Foundations of Behavioral Research, 4th ed., Harcourt College Publishers, 2000. Kim, Kwangmi Ko, "Advertising in Korean: International Challenges and Politics," in Advertising in Asia: Communication, Culture and Consumption, Katherine Toland Frith, Iowa State University Press: Ames, Iowa, 1996, pp.125-153. Korean Broadcasting and Advertising Corporation, "Opening: The Era of Comparative Advertising," in Korean Advertising information, Seoul: Korea, June 1999, URL - http://www.kobaco.co.kr/adv_info/199906/main2.htm Korean Overseas Information Service, A Handbook of Korea, Seoul: Korea, 1983. Krugman, Herbert E., "Why Three Exposures May Be Enough," Journal of Advertising Research, Dec. 1972, pp.11-15. Lamb, Charles W., William M. Pride, and Barbara A. Pletcher, "Taxonomy for comparative advertising research," Journal of Advertising, Vol.7, Winter 1978, pp.43~47. Lavidge, R. J. and G. A. Steiner, "Model for Predictive Measurements of Advertising Effectiveness," Journal of Marketing, Vol. 25, 1961, pp.59-62. Lyi, Dae-Ryong, A Study of the Effectiveness of Comparative Advertising, HanYang University, Seoul: Korea, 1988. McDougall, G. H. G., "Comparative Advertising: The Effect of Claim Type and Brand Loyalty," in James Leigh and Claude R. Martin (eds.), Current Issues and Research in Advertising, 1978, pp.39-52 Mitchell, Andrew and Jerry C. Olson, "Cognitive Effects of Advertising Repetition," in Advances in Consumer Research, Vol. 4, ed. W. D. Perreault, Atlanta, GA: Association for Consumer Research, pp.213-220. Muehling, Darrel D., "Comparative Advertising: The Influence of Attitude-toward-the-Ad on Brand Evaluation," Journal of Advertising, Vol.16, No.4, 1987, pp.43-49. , Jeffrey J. Stoltman, and Sanford Grossbart, "The Impact of Comparative Advertising on Levels of Message Involvement," Journal of Advertising, Vol.19, No.4, 1990, pp.41-50. Nunnally, Jum C., Psychometric theory, New York : McGraw-Hill, 2nd ed., 1978 Osgood, Cornelius, The Koreans and Their Culture, The Ronald Press Company, NY: New York, 1951. Palda, Kristian S., The Management of Cumulative Advertising Effects, Englewood, NJ: Prentice Hall, 1964. Pechmann, Cornelia and S. Ratneshwar, "The Use of Comparative Advertising for Brand Positioning: Association versus Differentiation," Journal of Consumer Research, Vol.18, Sep. 1991, pp.145~160. (a) and David Stewart, "The effects of comparative advertisement on attention, memory and purchase intentions," Journal of Consumer Research, Vol.17, Sep. 1990, pp.180~191. (b) , "Advertising Exposure: A Critical Review of Wearin and Wearout," Current Issues and Research in Advertising, Vol. 11, No. 2, 1988, pp.285-330. Petty, Richard E. and John T. Cacioppo, "The Elaboration Likelihood Model of Persuasion," in Advances in Experimental Social Psychology, Vol. 19, Leonard Berkowitz, ed. Orlando, FL: Academic Press, Inc., 1981, pp.123-205. , and D. Schumann, "Central and Peripheral Routes to Advertising Effectivenness: The Moderating Role of Involvement," Journal of Consumer Research, Vol. 10, Sep. 1983, p.135-146. , and M. Heesacker, " The Use of Rhetorical Questions in Persuasion: A Cognitive Response Analysis," Journal of Personality and Social Psychology, Vol. 40, Mar. 1981, pp.432-440. Prasad Kanty V., "Communication-Effectiveness of Comparative Advertising: A Laboratory Analysis," Journal of Marketing Research, Vol. 13, May 1976, pp.128-137. Pride, William M., Charles W. Lamb and Barbara A. Pletcher, "The Informativeness of Comparative Advertisements: An Empirical Investigation," Journal of Advertising, Vol.8, June 1979, pp.29~35. Sawyer, Alan and Scott Ward, "Carry-over Effects in Advertising Communication" Evidence and Hypotheses from Behavioral Science," In Darral Clarke (ed.) Cumulative Advertising Effects: Sources and Implications, Report No.77-111, Cambridge, MA: Marketing Science Institute, Sep. 1977, pp.71-170. Schumann, David, Richard Petty, and D. Scott Clemons, "Predicting the Effectiveness of Different Strategies of Advertising Variation: A Test of the Repetition-Variation Hypotheses," Journal of Consumer Research, Vol. 17, Sep. 1990, pp.192-202. Singh, Surenda N., Michael L. Rothchild, and Gilbert A. Churchill, Jr., "Recognition Versus Recall as Measures of Television Commercial Forgetting," Journal of Marketing Research, Vol. 25, Feb. 1988, pp.72-80. , Sanjay Mishra, Neeli Bendapudi, and Denise Linville, "Enhancing Memory of Television Commercial Through Message Spacing," Journal of Marketing Research, Vol. 31, Aug. 1994, pp.384-392. Shimp, Terence A. and David C. Dyer, "The Effects of Comparative Avertising Mediated by Market Position of Sponsoring Brand," Journal of Advertising, Vol. 7, No.3, 1978, pp.13-19. Swinyard, William R., "The interaction Between Comparative Advertisin and Copy Claim Variations," Journal of Marketing Research, Vol. 18, May 1981, pp.175-186. Wilkie, William L. and Paul Farris, "Comparison Advertising: Problems and Potential," Journal of Marketing, Vol.39, Oct. 1975, pp.7~15. Winter, Fredrick W., "A Laboratory Experiment of Individual Attitude Response to Advertising Exposure," Journal of Marketing Research, Vol. 10, May 1973, pp.130-140. Zajonc, R. B., "Novelty, Complexity, and Hedonic Value," in Perception and Psychophysics, Vol. 8, 1978, pp.279-286. Ads Exposed to Control Group (NCA- NCA -NCA) [--- ??? Graphic Goes Here ---] H/L "Notebook" - Now, ESN. High Speed and Power It enables you to utilize the potentials of the functions it has by 100 MHz high-speed performance. Excellent Screen and Sound Excellent multi-media functions with installed 4x CD-ROM, sound card, and speaker. [--- ??? Graphic Goes Here ---] H/L It's the Era of Notebooks High Speed and Power It enables you to utilize the potentials of the functions it has by 100 MHz high-speed performance. The Lightest, The Smallest A notebook should be comfortable to carry. Light (1.7kg) and Small (9.0x7.7x1.5) BrainTank [--- ??? Graphic Goes Here ---] H/L The Name of Notebook Only - ESN BrainTank. The Lightest, The Smallest A notebook should be comfortable to carry. Light (1.7kg) and Small (9.0x7.7x1.5) BrainTank Excellent Screen and Sound Excellent multi-media functions with installed 4x CD-ROM, sound card, and speaker. Ads Exposed to LOW Group (Low-Low-Low) [--- ??? Graphic Goes Here ---] H/L Compare the Performance Before Your Decision-Making. High Speed and Power It enables you to utilize the potentials of the functions it has by 100 MHz high-speed performance, which any other notebooks never reached. Excellent Screen and Sound Installed 4x CD-ROM, Sound card, and speaker make itself different from others, which claims their merits as convenience. [--- ??? Graphic Goes Here ---] H/L "The Best Notebook" It's ESN now. High Speed and Power It enables you to utilize the potentials of the functions it has by 100 MHz high-speed performance, which any other notebooks never reached. The Lightest, The Smallest A notebook should be comfortable to carry. The lightest (1.7kg) and the smallest (9.0x7.7x1.5) than any other notebooks - BrainTank [--- ??? Graphic Goes Here ---] H/L A New Concept of the Notebook - ESN BrainTank. The Lightest, The Smallest A notebook should be comfortable to carry. The lightest (1.7kg) and the smallest (9.0x7.7x1.5) than any other notebooks - BrainTank Excellent Screen and Sound Installed 4x CD-ROM, Sound card, and speaker make itself different from others, which claims their merits as convenience. Ads Exposed to MID-Group (Mid-Mid-Mid) [--- ??? Graphic Goes Here ---] H/L Compare to IBM Higher Speed It enables you to utilize the potentials of the functions it has by 100 MHz high-speed performance, which is faster than IBM (75MHz). More Various Functions Installed 4x CD-ROM, Sound card, and speaker make itself different from IBM, which claims their merits as convenience only. [--- ??? Graphic Goes Here ---] H/L IBM - It's no more the Best. Higher Speed It enables you to utilize the potentials of the functions it has by 100 MHz high-speed performance, which is faster than IBM (75MHz). Smaller! Lighter! A notebook should be comfortable to carry. More Convenient than IBM (2.0kg/ 9.7x7.9x2.0) - BrainTank (1.7kg/ 9.0x7.7x1.5) [--- ??? Graphic Goes Here ---] H/L ESN vs. IBM Smaller! Lighter! A notebook should be comfortable to carry. More Convenient than IBM (2.0kg/ 9.7x7.9x2.0) - BrainTank (1.7kg/ 9.0x7.7x1.5) More various Functions Installed 4x CD-ROM, Sound card, and speaker make itself different from IBM, which claims their merits as convenience only. Ads Exposed to HIGH Group (High-High-High) [--- ??? Graphic Goes Here ---] H/L Compare to IBM Function: ESN > IBM Price: ESN < IBM ESN BrainTank IBM ThinkPad CPU 486 DX4-100MHz 486 DX4-100MHz Memory 8mb 8mb HDD 810mb 810mb Display 10.4" TFT 10.4" TFT Video 1mb 1mb Sound 32bit/Speaker/Mic. 16bit CDROM X4 X2 Size 9.0x7.7x1.5 9.7x7.9x1.7 Weight 1.7 kg 2.0 kg Price 2,820,000 won 3,250,000 won [--- ??? Graphic Goes Here ---] H/L ESN vs. IBM Function: ESN > IBM Price: ESN < IBM ESN BrainTank IBM ThinkPad CPU 486 DX4-100MHz 486 DX4-100MHz Memory 8mb 8mb HDD 810mb 810mb Display 10.4" TFT 10.4" TFT Video 1mb 1mb Sound 32bit/Speaker/Mic. 16bit CDROM X4 X2 Size 9.0x7.7x1.5 9.7x7.9x1.7 Weight 1.7 kg 2.0 kg Price 2,820,000 won 3,250,000 won [--- ??? Graphic Goes Here ---] H/L IBM - It's no more the Best Function: ESN > IBM Price: ESN < IBM ESN BrainTank IBM ThinkPad CPU 486 DX4-100MHz 486 DX4-100MHz Memory 8mb 8mb HDD 810mb 810mb Display 10.4" TFT 10.4" TFT Video 1mb 1mb Sound 32bit/Speaker/Mic. 16bit CDROM X4 X2 Size 9.0x7.7x1.5 9.7x7.9x1.7 Weight 1.7 kg 2.0 kg Price 2,820,000 won 3,250,000 won Ads Exposed to Increasing Group (Low-Mid-High) [--- ??? Graphic Goes Here ---] H/L Compare the Performance Before Your Decision-Making. High Speed and Power It enables you to utilize the potentials of the functions it has by 100 MHz high-speed performance. Excellent Screen and Sound Installed 4x CD-ROM, Sound card, and speaker make itself different from others, which claims their merits as convenience. [--- ??? Graphic Goes Here ---] H/L IBM - It's no more the Best. Higher Speed With 100MHz high-speed performance, which is faster than IBM, it enables you to utilize greatly the potentials of the Notebook. Smaller! Lighter! A notebook should be comfortable to carry. BrainTank (1.7kg / 9.0x7.7x1.5) is convenient and lighter than IBM (2.0kg / 9.7x7.9x1.7) H/L ESN vs. IBM Function: ESN > IBM Price: ESN < IBM ESN BrainTank IBM ThinkPad CPU 486 DX4-100MHz 486 DX4-100MHz Memory 8mb 8mb HDD 810mb 810mb Display 10.4" TFT 10.4" TFT Video 1mb 1mb Sound 32bit/Speaker/Mic. 16bit CDROM X4 X2 Size 9.0x7.7x1.5 9.7x7.9x1.7 Weight 1.7 kg 2.0 kg Price 2,820,000 won 3,250,000 won APPENDIX Advertisements Exposed to; ù Control Group (NCA- NCA -NCA) ù LOW Group (Low-Low-Low) ù MID Group (Mid-Mid-Mid) ù HIGH Group (High-High-High) ù Increasing Group (Low-Mid-High)