Cross-cultural advertising communication: Visual imagery, brand familiarity, and brand recall Andrey Mikhailitchenko a , ⁎ , Rajshekhar (Raj) G. Javalgi b , 1 , Galina Mikhailitchenko c , 2 , Michel Laroche d , 3 a College of Business Administration, California State University, Sacramento, 6000 J St., Sacramento, CA 95819-6088, USA b Nance College of Business Administration, Cleveland State University, 1860 East 18th St., Cleveland, OH 44114, USA c Psychological Institute of Russian Education Academy, 9(4) Mokhovaya St., Moscow, 125009, Russia d John Molson School of Business, Concordia University, 1455 De Maisonneuve Blvd. West, Montreal, Quebec, Canada H3G 1M8 a b s t r a c t a r t i c l e i n f o Article history: Received 1 May 2007 Received in revised form 1 October 2007 Accepted 1 November 2007 Keywords: Advertising Cross-cultural Imagery Brand familiarity Brand recall The focus of this paper is to address the issue of visual imagery in cross-cultural consumer research. The authors investigate the relationship between visual imagery, brand familiarity, and brand claim recall in two distinct cultural environments — the U.S. and Russia. The paper consists of two studies that explored imagery potency from the point of view of memory-evoking effects. The results suggest that image-intensive tools generate different returns depending on the level of brand familiarity and cultural media. The research fi ndings may be of interest to marketing scholars studying cross-cultural consumer behavior as well as to practitioners operating in international advertising and global brand building. © 2008 Elsevier Inc. All rights reserved. 1. Introduction With the dynamic growth of cross-cultural marketing commu- nications, marketers give more consideration to the selection of visual imagery communication tools. They believe that visual imagery- intensive carriers such as TV, magazines, and product packages are also the most powerful ones in terms of in fl uencing brand recall. However, the ef fi ciency of this, as well as other visual imagery- evoking tools, differs depending on consumer memory-related factors, such as brand familiarity, product category experience, and degree of consumer involvement. Richardson (1969) de fi nes imagery as quasi-sensory or quasi- perceptual experiences that we are self-consciously aware of and that exist in the absence of stimulus conditions, producing their genuine sensory or perceptual counterparts. According to Lutz and Lutz (1977) , imagery is a mental event involving visualization of a concept or relationship. MacInnis and Price de fi ne imagery as a process ... “ by which sensory information is represented in working memory (1987, p. 473). ” Hence, imagery can be multisensory processing, which may lead to better recall of information ( Babin and Burns 1997 ). Imagery has been studied extensively in the cognitive psychology fi eld and has drawn increased interest among consumer researchers ( Babin and Burns, 1997; Andrews et al., 1994 ). The reliance on imagery has grown sharply over the recent decades for a number of reasons, including the drive for global marketing and development of international brands, such as Coca-Cola, Nike, Pizza Hut, Pepsi, and Mercedes ( Branthwaite, 2002 ). Therefore an investigation of the role of imagery in cross-cultural format is important as more and more U.S. and foreign multinationals are reaching consumers in foreign markets, especially in transition economies (e.g., Russia). Researchers (e.g., Cleveland and Laroche, 2007) note that interna- tional consumer research primarily focuses on one of two areas: the search for common groups of consumers with similar tastes and preferences across countries; and understanding consumer differ- ences from the perspective of cultural, social, economic, and market- ing programs. While the fi rst area emphasizes the importance of the emergence of global consumers, the second area suggests that successful marketing and communications strategies begin with cultural sensitivity — being tuned to the nuances of behavior of consumers in foreign markets. Competition for world markets and the increasing sophistication of foreign consumers have led to an increasing need for sophisticated advertising and communication strategies. The effects of the imagery in cross-cultural context are of practical relevance, especially countries such as Russia, which are embracing a market based economy. Russian consumers may hold entirely different views toward advertising than those held in the U.S., even though U.S.-based advertisers and brands are predominantly Journal of Business Research 62 (2009) 931 – 938 ⁎ Corresponding author. Tel.: +1 916 278 6578; fax: +1 916 278 4233. E-mail addresses: mikhaili@saclink.csus.edu (A. Mikhailitchenko), r.javalgi@csuohio.edu (R.(R.)G. Javalgi), galina.mikh@gmail.com (G. Mikhailitchenko), laroche@jmsb.concordia.ca (M. Laroche). 1 Tel.: +1 216 216 687 4789; fax: +1 216 687 9354. 2 Tel.: +7 495 202 9363. 3 Tel.: +1 514 848 2424x2942; fax: +1 514 848 4576. 0148-2963/$ – see front matter © 2008 Elsevier Inc. All rights reserved. doi: 10.1016/j.jbusres.2007.11.019 Contents lists available at ScienceDirect Journal of Business Research featured in Russia ( Andrews et al., 1994 ). Research is needed on cross- cultural understanding and applicability of advertising communica- tions to countries such as Russia. Companies that compete globally need to take cultural differences into consideration when making the decision regarding the advertising strategies applied in each country. The present study investigates the relationships between the brand familiarity, visual imagery of an ad, and the brand claim recall in two distinct cultural environments — the U.S. and Russia. These two cultures are chosen as they are highly distinct in terms of social and cultural values, economic and political conditions, and media habits. The marked differences between the U.S. and Russia help marketers to test cross-cultural impact on advertising communications. The paper concludes with an interpretation of the fi ndings and implications for cross-cultural advertising research. 2. Literature overview In the area of cognitive psychology, a great deal of research has been devoted to imagery. Various studies make a distinction between different types of imagery in terms of their sources, cognitive and affective associations, bases of raw stimuli, and avenues of informa- tion processing ( Richardson, 1969; Burns et al., 1993 ). Though imagery is believed to be multisensory (olfactory, auditory, tactile, taste), any considered type of imagery is related to a human's visual sensoric complex. Based on this categorization scheme, imagery-related studies operationalize ‘ imagery ’ and ‘ visual imagery ’ constructs as theoretical equivalents. 2.1. Research on imagery and advertising effectiveness relationship Rossiter and Percy (1980) , and LaBarbera et al. (1998) document memory recall to be higher if associated with more intensive visual imagery activity rather than under conditions that are believed to be less imagery stimulating. Multiple code theory ( Paivio, 1986 ) explains this effect by the fact that visual imagery creates multiple cues in human memory, and multiple retrieval processes associated with these cues increase the probability of recall. Based on this theoretical framework, some studies in marketing literature consider effect of imagery on brand attitude and brand claim recall ( Dobni and Zinkhan, 1990; Burns et al., 1993 ). The results of these studies demonstrate positive relationships between various dimensions of imagery (vivid- ness, concrete vs. abstract wording, instructions vs. no instructions to imagine, etc.) and the subsequent advertising effects (attitude toward the ad, attitude toward the brand, intention to buy, etc.). 2.2. Cross-cultural aspects The role of imagery in consumer perception and elaboration of information received in the process of international marketing communications has been a subject of the extensive scholarly and practical research within the last two decades. The distinctions in information processing, including its visual component, are caused by the fact that cultural environment is the complex set of beliefs, values, norms and attitudes acquired by consumers as part of their national heritage. Mueller (1992) and Harvey (1993) consider cultural environment to have a signi fi cant in fl uence on international market- ing communications strategy and corresponding choice of brand claim recall evoking communication tools. Some researchers tended to cluster the countries to determine the degree of international marketing communications standardization ( Sriram and Gopalakrishna, 1991; Katz and Lee, 1992 ). Sriram and Gopalakrishna (1991) identify six groups of countries and argued that communication strategies and imagery tactics could be attempted within each group by employing similar but not identical messages. The researchers fi gured out a number of environmental variables as important determinants of transferability of imagery communication, such as rate of economic growth, attitudes towards wealth and monetary gain, development and acceptance of international trade- marks, religious norms and beliefs, etc. (e.g. Dunn, 1976 ). Scholarly studies pay a great deal of attention to the psychographic and behavioral segmentation of image communication recipients in cross-border settings. Pollay (1986) and Green et al. (1975) recognize the U.S. advertising as heavily imagery-loaded, given its pervasiveness (9 min of TV advertising per hour, one of the largest imagery advertisers in the world). The research indicates that an American consumer has a higher degree of predisposition to perception of visual imagery and generates greater numbers of image-related responses than consumers in most of the other countries in the world ( Andrews and Lysonski, 1991 ). 3. Hypotheses The hypothesized link between brand familiarity, visual imagery, and brand claim recall is shown in Fig. 1 , and the hypothesized cross- cultural differences can be represented as shown in Fig. 2 3.1. Relationship between brand familiarity and brand claim recall Brand familiarity re fl ects the ‘ share of mind ’ of a given consumer attained to the particular brand and the extent of a consumer's direct and indirect experience with a brand ( Alba and Hutchinson, 1987; Kent and Allen, 1994 ). Campbell and Keller (2003) argue that brand familiarity is determined by strength of associations that the brand name evokes in consumer memory, and in this way it captures the consumer's brand attitude schemata. Their fi ndings demonstrate a main effect of brand familiarity on uncued brand recall such that familiar brands were better recalled than were unfamiliar brands. Fig. 1. The hypothesized model of relationship between brand familiarity, visual imagery of an ad, and brand claim recall. Fig. 2. The hypothesized interactions between culture and familiarity — recall and imagery-recall relationships. 932 A. Mikhailitchenko et al. / Journal of Business Research 62 (2009) 931 – 938 From a broader theoretic sense, not only the cited above signal detection and information organization theories, but also the recent research on message processing and message response (e.g., Campbell and Keller, 2003; Kent and Kellaris, 2001 ) provide the evidence of positive relationship between brand familiarity and brand claim recall. For instance, Kent and Kellaris (2001) fi nd that high levels of prior experience with a brand lead to the retention of stronger advertise- ment-brand links, making the attributes of previously familiar brands easier to recall. They posit that new attributes are linked more strongly to the nodes of highly familiar brands, which could ease the retrieval of advertisement claims. Based on the above, the following hypothesis is proposed. H1. A consumer's ability to recall brand information conveyed in an ad is related to his/her level of brand familiarity. 3.2. Relationship between visual imagery and brand claim recall The existing literature on cognitive psychology and consumer behavior provide the evidence in favor of better recall of information contained in imagery-intensive ads, in contrast to imagery non- intensive ones (e.g., Burns et al., 1993; Unnava and Agarwal, 1996; LaBarbera et al., 1998 ). While elaborating on cognitive mechanisms leading to this effect, Unnava and Agarwal (1996) state that when people process advertis- ing information that is high in image-provoking ability, it is anticipated that people will generate images prompted by that information. They conclude that these images are processed and integrated with the other information in the ad before being stored in the long-term memory. The literature studying the effect of word content of advertising on brand recall also documents the imagery-recall relationship ( Lutz and Lutz, 1977; Unnava and Burnkart, 1991 ). These studies demonstrate that high imagery words are remembered better than low imagery words and generate results in the formation of verbal and imaginal codes in memory. Based on this reasoning, we propose the following hypothesis. H2. A consumer's ability to recall brand information conveyed in an ad is related to the level of visual imagery of an ad. 3.3. Moderating role of visual imagery The consumer behavior and advertising communication research studies emphasize that information elaboration resources in the human mind are limited, and communication media are competing for them. Based on this view, when exposed to an ad for a familiar brand , consumers are likely to engage in relatively less extensive, more con fi rmation-based processing ( Keller, 1991; MacKenzie and Spreng, 1992 ). Familiarity can itself use cognitive capacity such that processing of a familiar, relative to an unfamiliar, stimulus is diminished ( Britton and Tesser, 1982 ). At the same time, the more extensive processing elicited by ads for unfamiliar brands increases the resource availability ( Cacioppo and Petty, 1979 ). Based on these fi ndings, the higher effect of imagery on the brand claim recall for the low level of brand familiarity is hypothesized: H3. The effect of visual imagery on brand claim recall is higher for the low level of brand familiarity, and is lower for the high level of brand familiarity. 3.4. Cross-cultural in fl uence The problems associated with communicating to consumers in diverse cultures pose the great creative challenges in devising advertising strategies. The cross-cultural advertising research litera- ture provides the evidence that the ad imagery ef fi ciency should differ from one cluster of countries to another ( Sriram and Gopalakrishna, 1991; Katz and Lee, 1992 ). The U.S. and Russia belong to different clusters that are caused by different cultural backgrounds and information processing patterns. According to Hofstede (1980, 1983) , the U.S. culture has higher scores on individualism, masculinity and long-term orientation, while Russia has higher power distance and uncertainty avoidance scores. Across all these dimensions, the U.S. and Russia are located far from each. The attitude towards the advertising in general ( Lutz, 1985 ), the theoretical and empirical link of which with individualism and masculinity in a culture was documented, also signi fi cantly differs for the two countries, making imagery advertising a more effective marketing tool in the U.S. than in Russia ( Mikhailitchenko and Whipple, 2006 ). Thus, the following hypothesis is proposed. H4. The relationship between ad imagery rating of an ad and brand information recall will be stronger for the U.S. than for Russia. 3.5. Interaction effects The relationship between brand familiarity and brand information recall may be different for the two countries as well. Laroche et al. (2002) document that this relationship works in different ways in mature and emerging markets. In mature, highly competitive markets, advertising may be functioning to maintain the status quo, while in emerging markets it is primarily a tool of creating brand value for newly established brands ( D'Souza and Rao, 1995 ). Therefore, information processing mechanisms is different for mature vs. emerging markets. In mature and highly competitive environments, it is likely that individuals process the advertised information based on established brand schemata, while in non- mature markets this schemata is usually absent. Respectively, brand familiarity in mature markets is usually associated with these schemata, while on emerging markets brand familiarity does not mean presence of an established network of associations with brand image ( Cacioppo and Petty, 1979 ). Based on this reasoning, it would be relevant to hypothesize that relationships between brand familiarity and brand claim recall will be stronger in mature rather than on emerging markets. Based on the above discussion, we propose the following hypothesis. H5. The relationship between brand familiarity and brand informa- tion recall will be stronger for the U.S. than for Russia. One of the culturally-oriented factors that contribute to the difference of imagery ads processing between the U.S. and Russian consumers is cross-cultural media habits. Extant literature demon- strates that people in the U.S. watch considerably more television and read more imagery-intensive print editions than people from Western Europe ( Green and Langeard, 1975 ) and Asia ( Lee and Tse, 1994 ). The same comparison would be relevant between the U.S. and Russia as well. In Russia people watch less TV, read more books, and are in a less degree exposed to ad carrying communication means than consumers in the U.S. ( Andrews et al., 1994 ). Recent literature on media habits in Russia demonstrates that in spite of the increased use of imagery- intensive carriers such as TV and Internet, especially among young populations, the Russian media environment is still characterized by the high role of books, newspapers and other written media ( Holak et al., 2007; Savel'eva, 2007 ). According to the market research organization NOP World, while an average U.S. media consumer spends 19 h per week on watching TV, and 5.7 h per week on reading, in Russia these indicators are 15 and 7.1 h per week respectively ( Baker, 2005 ). Based on the above, it would be logical to propose lower propensity of Russian consumers to advertising image elaboration 933 A. Mikhailitchenko et al. / Journal of Business Research 62 (2009) 931 – 938 than of their U.S. counterparts, and state interaction-related hypoth- esis as the following: H6. The moderating role of visual imagery on the relationship between brand familiarity and brand claim recall will be higher for the U.S. than Russia. 4. Sample Data were collected from two culturally diverse countries — the U.S. and Russia. The samples were drawn from university students in each country. The use of student samples for this study is justi fi ed by the fact that students have been regarded as one of fi ve important consumer segments both in the U.S. and Russia ( Andrews et al., 1994 ). In order to address the common concern about the use of student samples in cross-cultural studies, the homogeneity of respondents across two cultures (the U.S. and Russia) has been given special consideration ( Douglas and Craig, 2000 ). Both student samples represented equally sized and ranked business schools; they included approximately equal number of male and female respondents with age falling within the 20 to 30 years old range. 5. Study 1 (testing Hypotheses 1 – 3) 5.1. Design, measurement, and procedure For Study 1, the sample consisted of ninety six undergraduate students from two U.S. universities. The experiment was conducted in classroom settings in which each subject was randomly assigned to one of the treatments in 2 × 2 factorial between-subjects experimental design. Two conditions were manipulated — brand familiarity and imagery content of the ad. 5.1.1. Brand familiarity condition The two chocolate brands — American (A) and Russian (R) — both of which operate on the U.S. and Russian markets, were chosen for the experiment. While the Brand A one is the global brand and one of the market leaders in the U.S., the Brand R is being sold in local stores in some regions of the U.S. In contrast, Brand R is well-known as a market leader in Russia (approximately 30% market share), while Brand A is just one of the several imported brands with market share less than 1%. 5.1.2. Imagery condition Two ads, one consisting of picture and text, and the other one consisting of text only, were created for each brand. In order to exclude the in fl uence caused by the ad familiarity factor, completely new ads were generated for each of them. Each ad, for both of the brands, contained four attribute claims and one bene fi t claim. Among the four attribute claims, two were different for brands A and R, and two were the same. The bene fi t claims were different ( Table 1 ). 5.1.3. Procedure and measurement First, the students fi lled the questionnaire measuring the brand familiarity for brands they were assigned to. For brand familiarity measuring, Kent and Allen's (1994) scale in 7-point Likert format was applied. After that they were offered to examine the ads for one and a half minutes, and then rate the visual imagery level of ads by fi lling another questionnaire. For visual imagery measurement, Ellen and Bone's (1991) communications-evoked imagery processing scale was used. The measurement was made across three dimensions of imagery processing: vividness, quantity, and elaboration (adapted from Babin and Burns, 1997 ). After examining the ads, subjects were asked to complete a short “ Attitude towards advertising in general ” questionnaire that was used as an “ attention break ” between examining the ads and the next part of the experiment. Following this interpolated task, they then were given a surprise claim recall test on the ads. The recall period was kept within the limit of 2 min per ad. Subjects gave their responses in written; later these responses were coded and scored on 7-point scale, based on the number of correctly recalled attribute and bene fi t claims. 5.1.4. Stimulus design The choice of the product category (chocolate) was conditioned by the following factors. First, it should be familiar to the students, so that respondents would be able to form behavioral attitudes and visual images based on product category knowledge ( MacInnis and Price, 1987 ). Second, product category should be one where favorable visualizing experiences and brand claim recall could result in positive purchase intentions ( Rossiter and Percy, 1980 ). 5.2. Statistical analysis 5.2.1. Manipulation checks For both manipulated variables (familiar/unfamiliar brand and high/low imagery) manipulation checks were performed. Both for brand familiarity and imagery rating, descriptive statistics (mean values) of the responses to the corresponding parts of the ques- tionnaire were computed. The mean difference was signi fi cant for treatment conditions for both for brand familiarity ( F = 31.39, p b .001) and imagery rating ( F = 21.28, p b .001). 5.2.2. Convergent and discriminant validity of measurement In order to address convergent and discriminant validity issues, exploratory factor analysis was performed on items measuring the two independent variables discussed above. These two constructs were labeled respectively as FAMILIAR and IMAGE. The number of meaningful factors and number of items per factor were assessed with exploratory factor analysis. The factor analysis demonstrated the satisfactory internal con- sistency of the scale. The number of extracted factors (principal components rotation) with eigenvalues equal or greater than 1 was four (three dimensions of IMAGE and one for FAMILIAR). It should be noted here that there were three correlated dimensions of the IMAGE construct. Therefore, these three dimensions were collapsed into one dimension and labeled as “ IMAGE ” . Factor loading pattern for items included in the scale demonstrated a generally satisfactory fi t to the conceptual content of each factor. The con fi rmatory factor analysis (CFA) procedure was applied as well. The convergent validity issue was addressed by examining the signi fi cance of standardized factor loadings. All of them were on satisfactory level ( t -values from 8.8 to 18.9, all signi fi cant for p b .001). The discriminant validity was checked by performing a con fi dence interval test that involves calculating a con fi dence interval of plus or minus 2 standard errors around the covariances between the factors, and determining whether this interval includes 1.0 ( Anderson and Gerbing, 1988 ). 5.2.3. Measurement reliability For reliability assessment, Cronbach alpha coef fi cients were obtained. The reliability estimates were (.96 for familiarity and .77 for imagery) considered acceptable, based on Nunnally (1978) criteria. Table 1 Attribute and bene fi t claims for Brand A and Brand R. Brand A Brand R Attribute claims Roasted peanuts Raisins Caramel Hazelnuts Milk Milk Chocolate Chocolate Bene fi t claim Handles the hunger All natural 934 A. Mikhailitchenko et al. / Journal of Business Research 62 (2009) 931 – 938 5.3. Results A two-way ANOVA followed by post-hoc tests was used for testing the hypothesized main and interaction effects. As it follows from Table 2 , brand familiarity ( F = 69.44, p b .001) and imagery ( F = 18.77, p b .001) main effects as well as interaction effect ( F = 9.00, p b .01) were found to be signi fi cant. Student – Newman – Keuls and Waller – Duncan post-hoc tests were applied for evaluating the difference of RECALL means for all combina- tions of treatments. Both tests put High Familiarity (HF)/Low Image (LI) and HF/HI conditions in one homogeneous group, but at the same time put LF/LI and LF/HI conditions in different groups ( Table 3 ), supporting the hypothesized effect of higher imagery in fl uence in low rather than in high familiarity conditions (Hypothesis 3). Graphically the results of Study 1 are represented in Fig. 3 . They indicate at higher marginal in fl uence of imagery on lower rather than on higher level of brand familiarity. 6. Study 2 (testing Hypotheses 4 – 6) 6.1. Design, measurement, and procedure For study 2, the sample included eighty three undergraduate students from a U.S. university and one hundred nineteen students from a Russian university. Like in Study 1, the experiment was conducted in classroom settings in which each subject was randomly assigned to one of the treatments (HF/HI, HF/LI, LF/LI, and LF/HI); 2 × 2 × 2 factorial between — subjects experiment was designed with three manipulated conditions — a subject's country of domicile, brand familiarity and imagery content of the ad. The same chocolate brands, Brand A and Brand R, were used. For conducting the experiment in Russia the ad for both brands was translated into Russian with preserving exactly the same ad design and the same information for brand attributes and bene fi ts as in English. To ensure the conceptual equivalence of instructions and survey items in the two countries, a translation/back translation process was employed ( Andrews et al., 1994; Douglas and Craig, 2000 ). Translating survey instruments helps ensure that items and response formats have identical meanings across cultures. Otherwise, “ cross-national differences in scale means might be due to differences between countries on the underlying construct or due to systematic biases in the way people from different countries respond to certain items ” ( Steenkamp and Baumgartner, 1998 ). 6.1.1. Brand attitude as covariate Except country of domicile condition, another difference with Study 1 was that Attitude towards brand score ( A B ) was measured as well. The four-item seven-point differential scale was used for that purpose (Appendix A), and the items were averaged (Cronbach's alpha = .92). The major reason for measuring brand attitude was the premise that the better brand attitude might predispose subjects to more detailed recall of an ad conveyed brand information (e.g., Patzer, 1991; Coulter and Murphy, 1994 ). Once brand attitude was formed before the experiment as a result of previous consumer experience, its in fl uence on brand recall, unless is put under control in statistical analysis, could question the results. The inclusion of A B variable in the model appears to be relevant for the cross culture — formatted study, taking into account the hypothesized culture-related differences of imagery impact on brand recall. The procedure was the same as in Study 1, with the exception that after fi lling in the brand recall related part of the questionnaire they also answered brand attitude measuring questions. 6.2. Statistical analysis 6.2.1. Manipulation checks Like in Study 1, manipulation checks were performed for the independent variables (familiar/unfamiliar brand and high/low imagery). The mean difference for IMAGE was signi fi cant for picture vs. non-picture condition ( F = 30.31, p b .001). However, the difference for FAMILIAR variable was found to be signi fi cant for the U.S. sub- sample ( F = 25.50, p b .001), but insigni fi cant for the Russian sample ( F = .734, p b .393). In order to create a categorical FAMILIAR variable for subsequent ANCOVA procedure 8 observations out of the initial 127 ones were removed from the analysis, and the delineation between HF and LF conditions became possible ( F = 3.73, p b .05). Unlike for Study 1 and the U.S. sub-sample for Study 2, sub-sample for HF condition (as well as LF one) included both brands: Brand A and Brand R. 6.2.2. Validity and reliability of measurement The same procedures for checking discriminant and convergent validity and reliability of measurement as in Study 1 were applied. The exploratory factor analysis was performed on three instead of two variables (included brand attitude- labeled as ATTITUDE). The number of meaningful factors with eigen values more than 1 was fi ve (in compliance with 3 dimensions of IMAGE plus FAMILIAR and ATTITUDE). Factor loading pattern demonstrated satisfactory model fi t ( t -values from 7.4 to 19.1, signi fi cant for p b .001). The Phi- Table 2 ANOVA results. Dependent variable: RECALL Source Type III sum of squares df Mean square F Sig. Corrected model 43.750 3 14.583 32.407 .000 Intercept 174.050 1 174.050 386.778 .000 FAMILIAR 31.250 1 31.250 69.444 .000 IMAGE 8.450 1 8.450 18.778 .001 FAMILIAR ⁎ IMAGE 4.050 1 4.050 9.000 .008 Error 7.200 92 .450 Total 225.000 96 Corrected total 50.950 95 R squared = .859 (adjusted R squared = .832). Table 3 Results of Student – Newman – Keuls and Waller – Duncan tests. Treatment N Subset 1 2 3 Student – Newman – Keuls LF/LI 25 2,0241 LF/HI 23 2,8511 HF/LI 24 4,0682 HF/HI 24 4,3396 Waller – Duncan LF/LI 25 2,0241 LF/HI 23 2,8511 HF/LI 24 4,0682 HF/HI 24 4,3396 Fig. 3. Recall mean scores for combinations of treatments. 935 A. Mikhailitchenko et al. / Journal of Business Research 62 (2009) 931 – 938 Correlation indexes ranged from .2401 to .7494 with standard errors from .0611 to .2721. The inclusion of 1.0 in con fi dence interval took place only between factors measuring the dimensions of IMAGE construct. Cronbach alpha coef fi cients were .88 for familiarity, .80 for imagery, and .92 for brand attitude. The issue of cross-cultural stability of measurement was addressed as well. It was done by running separate exploratory factor analysis models for each of two countries and comparing results. The factorial similarity between the cultural groups within the sample was evaluated. This procedure of establishing cross-cultural (cross- sample) stability of a scale was applied in a considerable number of studies (e.g., Ommundsen et al., 2002; Lubatkin et al., 1998 ). For both countries, examination of the scree plots revealed fi ve major factors. Inspection of the items that loaded on these factors revealed that the dimensional structure of constructs had been successfully recaptured across two cultural groups. Next, the factor loadings of American and Russian groups on these dimensions were compared by means of Pearson product-moment correlations and Tucker's coef fi cient of congruence. These coef fi cients ranged from .95 to .99 that provided the ground for concluding that the factorial structure is quite comparable in cross-cultural settings and the two subsamples can be combined for subsequent analysis. 6.3. Results An ANCOVA with COUNTRY, IMAGE, and FAMILIAR as independent variables, RECALL as dependent variable, and ATTITUDE as covariate was used for testing the hypotheses. The results obtained so far are depicted in Table 4 The covariate (brand attitude score) was signi fi cant ( F = 5.618, p b .019). The main effects of imagery rating ( F = 8.969, p b .01) and brand familiarity ( F = 226.16, p b .01) on brand recall were signi fi cant even after removing the covariate's effect. Not only in terms of IMAGE and FAMILIAR main effect, but also their interaction effect ( F = 10.893, p b .01), the Study 2 replicated the results obtained in Study 1. The empirical fi nding about higher imagery effect under low rather than high familiarity conditions was supported in Study 2 as well. The signi fi cant 3-way interaction between country of domicile, brand familiarity, and imagery rating ( F = 7.057, p b .01) demonstrates different patterns of marginal in fl uence of the imagery on brand claim recall for high familiarity vs. low familiarity conditions in two cultures (Hypothesis 6). While in the U.S. imagery has a much higher contribution to brand recall on low levels of brand familiarity, in Russia the difference of this contribution for HF and LF conditions is much less substantial ( Table 5 ). Graphically, the most important results of Study 2 can be represented by a graph illustrating the difference in marginal contribution of imagery to brand recall score on high versus low levels of brand familiarity, compared for the U.S. and Russian sub- samples ( Fig. 4 ). The culture-related differences in imagery perception (Hypothesis 4) were supported by the signi fi cance of IMAGE ⁎ COUNTRY interac- tion effect ( F = 4.333, p b .05). In contrast, the hypothesized difference of brand familiarity in fl uence on brand recall (Hypothesis 5) was not supported by the study results ( F = .961. p b .328). The failure to obtain Table 4 ANCOVA results. Source Type III sum of squares df Mean square F Sig. Corrected model 216.150 8 27.019 37.439 .000 Intercept 46.294 1 46.294 64.149 .000 ATTITUDE 4.054 1 4.054 5.618 .019 ⁎ IMAGE 6.472 1 6.472 8.969 .003 ⁎⁎ COUNTRY 16.989 1 16.989 23.542 .000 ⁎⁎ FAMILIAR 163.209 1 163.209 226.156 .000 ⁎⁎ IMAGE ⁎ COUNTRY 3.127 1 3.127 4.333 .039 ⁎ IMAGE ⁎ FAMILIAR 7.861 1 7.861 10.893 .001 ⁎⁎ COUNTRY ⁎ FAMILIAR .693 1 .693 .961 .328 IMAGE ⁎ COUNTRY ⁎ FAMILIAR 5.093 1 5.093 7.057 .009 ⁎⁎ Error 139.281 193 .722 Total 2611.000 202 Corrected total 355.431 201 ⁎ Signi fi cant at p b .05. ⁎⁎ Signi fi cant at p b .01. Table 5 Means and standard deviations of brand claim recall for the combinations of treatments. Dependent variable: RECALL Imagery Country Brand familiarity Mean Std. deviation N Low USA Low 1.333 0.963 20 High 4.038 1.148 20 Total 2.740 1.724 40 RUSSIA Low 2.714 0.460 23 High 4.444 0.616 31 Total 3.391 1.000 54 Total Low 2.077 1.007 43 High 4.205 0.978 51 Total 3.052 1.454 94 High USA Low 2.800 1.361 22 High 4.000 0.894 21 Total 3.415 1.284 43 RUSSIA Low 2.875 0.751 27 High 4.545 0.506 38 Total 3.723 1.053 65 Total Low 2.846 1.017 49 High 4.333 0.727 44 Total 3.604 1.152 93 Total USA Low 2.000 1.364 42 High 4.021 1.032 41 Total 3.044 1.570 83 RUSSIA Low 2.800 0.632 65 High 4.510 0.543 54 Total 3.586 1.040 119 Total Low 2.462 1.079 107 High 4.276 0.847 95 Total 3.342 1.330 202 Fig. 4. Recall mean scores for combinations of treatments (for the U.S. and Russian sub- samples). Table 6 Results of hypotheses testing. Hypotheses Study 1 Study 2 Main effects on brand claim recall H1 (brand familiarity) Supported at p b .001 Supported at p b .001 H2 (ad imagery rating) Supported at p b .001 Supported at p b .003 Interaction effects H3 (familiarity ⁎ imagery) Supported at p b .008 Supported at p b .001 H4 (imagery ⁎ country) Supported at p b .05 H5 (familiarity ⁎ country) Not supported H6 (familiarity ⁎ country ⁎ imagery) Supported at p b .01 936 A. Mikhailitchenko et al. / Journal of Business Research 62 (2009) 931 – 938 the support for this hypothesis may mean that, in contrast to imagery, brand familiarity's impact on brand recall does not signi fi cantly differ across cultures, at least within the applied sample frame (USA vs. Russia). To summarize, Studies 1 and 2 provided the signi fi cant support for hypothesized brand familiarity and imagery rating effects on brand recall ( Table 6 ). Their interaction (higher contribution of imagery to brand claim recall on low rather than on high levels of brand familiarity) was supported by both studies as well. The signi fi cant differences of imagery in fl uence on brand recall score was revealed by the signi fi cance of corresponding interactions in ANCOVA model (Study 2). The difference of brand familiarity's in fl uence on brand recall across two cultures was not supported by the study ( Table 6 ). 7. Discussion This study addressed the issue of imagery potency from the point of view of memory-evoking effects under different levels of brand familiarity within different cultural contexts. The results document differential impact of the imagery on brand recall: it is higher for low- and lower for high-familiarity conditions. In cross-cultural settings, on Russian market brand recall — evoking ability of imagery found out to be less than in the U.S. From a theoretical perspective, the fi ndings of the study are consistent with earlier research investigating the effects of brand familiarity and imagery communication on brand recall ( Babin and Burns, 1997; Burns et al., 1993; Branthwaite, 2002 ) as well as with cross-cultural advertising and consumer behavior studies ( Andrews and Lysonski, 1991; Andrews et al., 1994; Mueller, 1992 ). At the same time, it builds on the existing research by providing a more nuanced view of when and how diminishing effects of imagery are likely to occur. The fi ndings reported in this study attract attention to the issue of information processing resources that exist in the consumer mind. The high level of brand familiarity creates the strong brand-related schema that can use the cognitive capacity and interfere in the elaboration of the new image-evoked information. In contrast, the absence (or weakness) of such a schema for unfamiliar brands increases the availability of cognitive resources needed for image elaborating activity. The literature both in social cognition ( Carlston, 1980; Lingle and Ostrom, 1979 ) and consumer behavior ( Kardes, 1986; Britton and Tesser, 1982 ) developed this theoretical approach. The results of this study provide additional evidence for its further development and operationalization. The study also demonstrates that the process of elaboration of image-initiated information is strongly in fl uenced by social media. The need for image processing appears to be cultural phenomena that are in fl uenced by traditions of the country, its media habits, and dominating lifestyle. The strong cultural predisposition to reading books and newspapers create in less need for visualization, while media habits oriented toward watching TV and reading colorful magazines create image-based information elaboration patterns. The associative links evoked by images (e.g., milk chocolate, peanuts, raisins) result in higher attribute recalling abilities for consumers from image-intensive media environm