Utilitarian, Aesthetic, and Familiarity Responses to Verbal Versus Visual Advertisements

ABSTRACT - Our objective was to investigate consumers' perceptions of visual versus verbal advertisements along three dimensions-functional utility, aesthetic value, and familiarity. Based upon prior research and theory, it was hypothesized that verbal advertisements would be perceived as more utilitarian, whereas visual advertisements would be perceived as more aesthetic and familiar. The first and last hypotheses were consistently confirmed; the second hypothesis garnered mixed support.

Citation:

Elizabeth C. Hirschman and Michael R. Solomon (1984) ,"Utilitarian, Aesthetic, and Familiarity Responses to Verbal Versus Visual Advertisements", in NA - Advances in Consumer Research Volume 11, eds. Thomas C. Kinnear, Provo, UT : Association for Consumer Research, Pages: 426-431.

Advances in Consumer Research Volume 11, 1984      Pages 426-431

UTILITARIAN, AESTHETIC, AND FAMILIARITY RESPONSES TO VERBAL VERSUS VISUAL ADVERTISEMENTS

Elizabeth C. Hirschman, New York University [Associate Professor or Marketing]

Michael R. Solomon, New York University [Assistant Professor of Marketing and Associate Director, Institute of Retail Management]

ABSTRACT -

Our objective was to investigate consumers' perceptions of visual versus verbal advertisements along three dimensions-functional utility, aesthetic value, and familiarity. Based upon prior research and theory, it was hypothesized that verbal advertisements would be perceived as more utilitarian, whereas visual advertisements would be perceived as more aesthetic and familiar. The first and last hypotheses were consistently confirmed; the second hypothesis garnered mixed support.

INTRODUCTION

Beginning with the early work by Krugman (1971), researchers have become increasingly aware that consumers' involvement with and responses to advertising messages may be mediated not only by the information content of the message, but by advertisement format as well (cf. Appel, Weinstein and Weinstein 1979; Weinstein, Appel and Weinstein 1980). More specifically, researchers have examined differences between verbal message formats and visual message formats in terms of their effects on information processing, recall, and beliefs concerning the product (cf. Moore and Hutchinson 1983; Shimp 1981).

A major goal of the present research is to extent research in this area by emphasizing a potentially important distinction. While many studies have focused on the comparative effects of advertisement format, they have not considered the possibility of a confounding between information content of an ad and format of an ad. When format is varied, the information transmitted can vary as well. As a result, it is possible to conclude that format manipulations affect beliefs, recall, etc. when in fact such manipulations cause consumers to process a different set of information as well. For example, it is possible that visual ads impart more information ('a picture is worth a thousand words") than do print ads or that print ads are assumed to be more 'credible" and hence are better attended to. Thus, it was deemed important to recognize the orthogonal nature of presentation content and mode of presentation (cf. Solomon, Drenan and Insko 1981). In the present study, mode of presentation is varied (verbal versus visual). but information content is held constant.

In many of the previous studies, the format effects investigated have been print (e.g., magazines) versus televised (e.g., television) vehicles for advertising transmission (Appel, et al. 1979; Rossiter 1981; Weinstein, et al. 1980). These studies used the left brain/right brain metaphor as their guiding paradigm, and brain wave measurements as the criterion variable (Hansen 1981; Rossiter 1981). Another study (Golden and Johnson 1983) contrasted "thinking" versus 'feeling" television commercials in terms or their affect, informativeness, usefulness, believability and ability to stimulate purchase of the product.

Several other studies have utilized print advertisements as stimuli for investigating information content effects on consumers' responses to advertising (HolbrooK 1978; Lutz and Lutz 1977; Mitchell and Olson 1981, Srull 1983), but in very few instances has content format (i.e., visual versus verbal) been varied as well. In the study by Holbrook (1978), for example, verbal message content was manipulated--using rational versus emotional language-and the resulting effects on consumers attitude and intention to purchase were monitored. In the Srull (1983) study, both advertisement copy and subject affect were manipulated in three affective conditions--positive, negative, and neutral. Subjects' recall and attitude toward the product were measured using free response and semantic scaling techniques. In the Lutz and Lutz study (1977), interactive imagery (in which the brand name and a visual image were integrated) was compared to non-interactive imagery (in which brand name and visual image were discrete elements) and found to be superior in stimulating subjects' recall.

The Mitchell and Olson study (1981) utilized four print advertisements. One contained a brand name headline and an explicit product claim, but no picture (i.e., it was verbal only); the three remaining advertisements contained a brand name headline and picture, but no written copy (i.e., they were intended to be visual only). Subjects' beliefs regarding product attributes, affect toward the advertisement and product, and intention to purchase were measured using bipolar scaled adjectives.

Although the foregoing studies have examined various aspects of advertising message content and/or message format vis a vis a range of dependent measures, there are still several unexplored areas. In none of the studies cited, for example, was an attempt made to measure subjects' perceptions of an advertisement for (1) the same product using either (2) an all-visual or all-verbal format of (3) an actual advertisement. In the Mitchell and Olson (1981) study the advertisements were artificially developed for the purpose of examining alternative models of attitude structure. in the Golden and Johnson (1983) study, conversely, actual advertisements were used to represent ;thinking versus "feeling' conditions, but in each case different products were represented--leading to product-specific interactions. In the Srull (1983) research actual advertisements were used, but they consisted only of verbal (i.e., copy) information; no visual component was present. In the Holbrook study (1978), verbal information was manipulated in artificial advertisements and the visual component was not manipulated.

PURPOSE

It is our research objective, therefore, to assess subjects' perceptions of advertisements which meet the three conditions cited above: (1) the same product is used in (2) an all-visual or all-verbal presentation format (3) derived from an actual advertisement. In this design the confounding factors of product, format, and artificiality of stimulus are all brought under greater control--leading to a more direct and uncontaminated assessment of advertisement format effects on consumer perceptions.

Criteria

Prior research on this subject has utilized a wide variety of dependent measures. One stream-of investigation has been based on physiological indicators of consumer response to advertisements, most notably brain waves representing hemispheric activation (c.f. Appel et al 1977; Krugman 1971; Weinstein et al 1980). Despite their semblance of empirical soundness, however, these measures have been strongly criticized as invalid and subject to misinterpretation when applied to advertising response (Ray and Batra -1983; Rossiter 1981).

A second research tradition has utilized self-report semantic scales to measure a cross-section of perceptual criteria--credibility and believability (Golden and Johnson 1983; Holbrook 1978), affective-emotional arousal (Golden and Johnson 1983; Holbrook 1978; Lutz, Mackenzie and Belch 1983; Mitchell and Olson 1981; Srull 1983), informativeness, utility, factualness, and rationality (Golden and Johnson 1983; Holbrook 1978), content recall (Lutz, Mackenzie and Belch 1983; Srull 1983), and intent to purchase (Golden and Johnson 1983; Holbrook 1978; Mitchell and Olson; Srull 1993).

In the present research we utilize one perceptual dimension directly derived from this previous research--utilitarian/ rational, a second dimension which is derived from, but extends beyond, prior research--aesthetic/emotional, and a third dimension which to our knowledge has not been previously investigated within this context--familiarity. Ali three dimensions are loosely based on learning theory in that they involve learned associations between mote of presentation and information content. While the first two dimensions are influenced by social learning (socialization) theory, the predicted familiarity effects are based upon psychological learning theory findings on imagery and semantic processing. These three dimensions were all measured using self-report semantically anchored scales, which is perhaps the dominant operational mode in this research area.

The rationale, measurement, and hypotheses for each are described below:

Utilitarian/Rational

As noted earlier, several studies have used the utilitarian/rational dimension as a criterion of advertising response. Yet none of these studies has linked this dimension directly to an all-visual or all-verbal format. It was our contention that all-verbal print advertisements would be perceived as more utilitarian/rational than all- visual print advertisements for the same product. The reasoning to support this contention was that the U.S. cultural emphasis on reading, literacy, and education as dominant modes of rationality and logic, would cause verbal (i.e., written) presentations to be perceived as more utilitarian and rational than visual (i.e., picture) presentations (Polanyi and Prosch 1976). Thus, in much the same way that an all-text book is viewed popularly as more scholarly and serious than an all-picture book, we believe that a perception of greater rationality will adhere to an all-verbal print advertisement than to one that is all visual. [Note that this claim is not based on the absence of affective or emotion-laden words in the text of the all-verbal advertisement, but rather simply on the actual textual nature of the format, itself.]

The utilitarian-rational scale consisted of four adjective pairs: logical/not logical, educational/not educational, informative/not informative, and factual/not factual. The positive form of these adjectives is generally accepted as typifying rational processing of information (e g. t Holbrook 1978; Golden and Johnson 1983; Osgood, Suci, Tannenbaum, 1957). .

The formal statement of the hypothesis for this dimension was:

H1: Subjects viewing all-verbal print advertisements will rate them as being more utilitarian-rational than subjects viewing all-visual print advertisements for the same product.

Aesthetic/Emotional

Several prior studies have investigated the affective and/ or emotional aspects of advertising effects. in the present research, we extended the context of this dimension to include aesthetic response to a stimulus, as well as emotional response. As Holbrook and Zirlin (19&3) nave noted, aesthetic response to a stimulus is suffused with emotion, but additionally extends beyond emotion to include evaluative reactions to an object (e.g.. as being beautiful or sublime).

It was our contention that in U.S. culture, visual images (e.g., sculptures, paintings, graphics) are more generally regarded as aesthetic objects than are text passages (Holbrook and Zirlin 1583; Polanyi and Prosch 1576). This we believed is a common, socialized response tendency that would apply to advertisement components just as readily as to paintings in a museum. We believed, in effect, that consumers are culturally conditioned to equate visual images with art, beauty, and emotional attraction; while concurrently associating printed words with knowledge, logic, and 'actuality. To provide another cultural metaphor, paintings and graphics are displayed in art museums and galleries (i.e., collections of aesthetic objects), whereas books and printed periodicals are housed in libraries (i.e., collections of knowledge and information). [A reviewer has raised the issue of whether utilitarian-rational and aesthetic-emotional responses are opposite ends of the same continuum. By contrast we propose that they are orthogonal dimensions--independent of one another and ranging from high to low along separate continua. This posture reflects the reasoning of Holbrook and Zirlin (1983) and is analogous to the modern conceptualization of masculine and feminine sex roles as orthogonal dimensions, rather than opposites. A stimulus such as a car or house may possess both utilitarian and aesthetic features. One dimension does not necessarily decline as the other increases; rather they vary independently of one another (e.g., a house by Frank Lloyd Wright, a Honda Prelude).]

The aesthetic-emotional scale consisted of four adjective pairs: attractive/not attractive, desirable/not desirable, arousing/not arousing, and beautiful/not beautiful. Affirmative responses to adjectives of this type are characteristically regarded as aesthetic/emotional in content within our culture (e.g., Holbrook 1978; Osgood, Suci, Tannenbaum, 1957). Therefore, it was hypothesized that because their aesthetic sense would be activated by visual imagery, subjects viewing all-visual print advertisements would rate them as being more attractive, desirable, arousing, and beautiful than subjects viewing all-verbal print advertisements. More specifically:

H2: Subjects viewing all-visual print advertisements will rate them more highly on the aesthetic-emotional dimension than subjects viewing all verbal print advertisements for the same product.

Familiarity

To our knowledge, no prior studies reported in the advertising and marketing literatures have used stimulus familiarity as a criterion for comparing subjects t responses across unfamiliar (i.e , never viewed before) advertisements. It is our contention that visual images will be perceived as being more familiar than verbal descriptions or the same (unfamiliar) subject matter. There are several rationales for this. First, the psychological literature has indicated consistently that the learning of visual images (i.e., pictures) occurs more readily than the verbal counterparts of these same stimuli (c . f ., Bower 1970; Childers and Houston 1983; Paivio 1969; Zajonc 1980). These and other studies suggest that image processing is easier and more rapid than verbal processing. Although alternative explanations have been advanced for this phenomenon (c f., Paivio 1971; Ray and Batra 1983; Springer and Deutsch 1981), one direct implication is that because visual stimuli are more easily and rapidly comprehended than verbal stimuli, they may be perceived as less novel, less difficult, and less "uncomfortable" than verbal stimuli on the same topic. Thus, we believe, a pictorial image of an (unfamiliar) advertised product will be perceived as more familiar than a corresponding verbal product description. Even though the contents of both advertisement formats are equally unknown to the consumer, the visual format, we contend, will create the perception of greater familiarity:

H3: Subjects viewing all-visual print advertisements will rate them as being more familiar than subjects viewing all-verbal print advertisements for the same product.

The familiarity scale consisted of two adjective pairs: known to me/not known to me and familiar/not familiar.

Scaling Format

The adjective pairs composing the three criterion dimensions were used as anchor points on seven-point rating scales, and presented in randomized order. A "7" was used to indicate the positive end of each adjective pair and a "1" to indicate the absence end of the pair. It should be noted that the semantic structure of the scales represented a continuum from high presence of a characteristic to its complete absence and not high presence of a characteristic to high presence of its opposite. For example, the scale ran from "beautiful" to "not beautiful," as opposed to from "beautiful" to "ugly". This was because the identification of accurate semantic opposites for the characteristics under investigation (e.g., arousing, factual) proved quite difficult. Rather than anchor the continua with inaccurate antonyms, it was believed more valid to create continua that extended from the presence of a characteristic to its nonexistence, e.g., factual to not factual.

In the present study, each subject responded to each of the three sets of dimensional scales for two advertisements-that is, an advertisement was presented, the three sets of scales were filled out, a second advertisement was presented, and a second series of scales was completed. The alpha reliability coefficients are presented in Table 1. Each criterion is listed twice to represent measurements taken for each of the two advertisements. They ranged from .76 to .91, indicating a good to very good level of internal consistency. [The three scales also exhibited adequate levels of discriminant validity. Their intercorrelations were as follows: Util/Rat (1+2) by Aesth/Emot (1+2) r = .38, Util/Rat (1+2) by Famil (1+2) r = .22, Aesth/Emot (1+2) by Famil (1+2) r = .35.]

Advertising Stimuli

Selection of stimuli appropriate for the task appeared superficially straightforward, but actually involved several difficult issues. First, we desired to use real advertisements in order to enhance the external validity of the study (Srull 1983). However, this can create problems of subject familiarity with the product, which may confound the results. Second, we desired to use advertisements whose visual content corresponded with their verbal content; that is, which provided similar messages via picture and text. [Note that total isomorphism between picture and text is neither required nor expected. The fact that picture and text were hypothesized to create somewhat different perceptions in respondents negates the possibility of complete meaning congruence between these two advertisement components.]

The first advertisement was taken from Connoisseur magazine, a British publication that deals with antiques. The advertisement depicted a statue of the Greek goddess Fortuna. This advertisement had two unique advantages. First, it was highly unlikely that any of the subjects (college students and their social contacts) would have seen it previously. Second, the verbal description of the product and its visual depiction are guaranteed highly accurate by norms operating within the antiquities industry. It is considered highly improper to falsely portray an antiquity either in picture or verbal description to potential clients. Hence, we believed that the picture and text in the advertisement attempted to provide as accurate--and equivalent--information regarding the product as was possible. The second advertisement was taken from a year-old issue of Architectural Digest, a magazine specializing in interior decoration for the very affluent; the product was a chrome, glass, and marble dining table. It was believed unlikely that the subjects would have been exposed to this advertisement; it was judged to possess generally consistent verbal and visual content, based upon the expert opinion of two advertising copywriters familiar with this product category. [Since the objective of the study was to investigate subjects' perceptions of the stimuli, and not to predict actual purchasing behavior, the fact that the subjects may not be purchasers of these specific products is inconsequential to the experiment. Further, their probable lack of familiarity with either brand strengthened the test o the third hypothesis -- i.e., that unfamiliar pictures would seem more familiar than]

Subjects were provided with a packet of stapled pages containing one format only of each of the two advertisements. The order of the two advertisements was not varied across subjects. While this may have the potential for creating some order effects, it was not believed that this would compromise the objectives of the study. Further, by maintaining the same stimulus order throughout, we were able to detect effects specific to a particular advertisement versus those which generalized across advertisements. The semantic differential scales were placed behind each advertisement. Hence, the criteria of experimental interest were replicated within the study design.

One additional issue with respect to the advertising stimuli also deserves mention. On both the all-verbal and all-visual treatments of each advertisement, the sponsor's name appeared exactly as it had in the original advertisement. Although a "pure" all-visual stimulus would contain no text, we believed the inclusion of the sponsor's name was appropriate for the following reasons. First, to enhance the study's internal validity it was necessary to create the illusion of an actual advertisement, even though the message consisted of only a picture. The inclusion of the sponsor's name, we believed, helped to create that illusion (Mitchell and Olson 1981). Second, in both of the advertisements, the sponsor's name was foreign and unlikely to evolve any semantic associations (Srull 1983). Thus, the inclusion of the sponsor's name would seem, on balance, to be constructive and not liable to seriously impugn the format purity of the stimuli. [It is not clear if a sponsor's name which appears in the form of a stylized logo is, in fact, mentally processed as text. Rather, it has been argued that the cue configuration is perceived holistically and thus constitutes an image (Rossiter 1980).]

Control of Moderating Factors

One shortcoming of several studies we reviewed was that little, if any, consideration was given to moderating factors which might influence consumers' response patterns independent of advertisement format, or which may interact with advertisement format to influence their response patterns. Two exceptions to this pattern were the studies by Srull (1983) and Golden and Johnson (1983). Srull manipulated the mood of his subjects both in valence (i.e., positive, negative) and intensity (i.e., moderate, high). These affective manipulations were found to significantly influence ad recall and evaluation. In the Golden and Johnson study, subjects' a priori preferences for visual or auditory stimuli were used as covariates, but were not found to affect responses to "thinking" versus "feeling" commercials.

In addition to these variables, there are several other factors that logically would appear to be linked to a subject' reaction to visual versus verbal advertisement format. For example, some individuals have a pronounced ability to generate internal imagery, whereas others lack this ability (Richardson 1969). Some individuals seek high levels of emotional involvement in their consumption activities, whereas others avoid it (Holbrook and Zirlin 1983). Individuals have differential tendencies to seek cognitive and/ or sensory stimulation (Pearson 1970). Finally, individual may tend to respond to stimuli in an holistic/projective fashion or in an instrumental/analytical manner (Ornstein 1977). In each of these instances, consumers may display different response patterns to verbal or visual stimuli due to preexisting tendencies/abilities. It is possible that these factors may either interact with stimulus content or act independently of stimulus content to affect the criteria under examination here. Hence, their effects should be explicitly controlled for within the research design. To accomplish this, subjects were administered the following scales prior to undertaking the advertisement response task. [Details on moderator variable measurement and validity are available from the authors. Space constraints prohibit their inclusion here.]

Imagery. It was believed that high or low levels of imagery ability might react with advertising format (visual vs. verbal) to affect the subject's response. It was specifically hypothesized that degree of imagery ability would polarize the sample in that high-imagery subjects would be more receptive to visual stimuli and thus exhibit higher ratings on the aesthetic/emotional and familiarity dimensions.

Emotional Involvement. It was hypothesized that subjects predisposition toward emotional involvement would moderate responses, in that visual stimuli are more engaging of a sensory channel as opposed to the linear processing required by verbal stimuli. Thus, the moderating effects of this control variable, if any, were predicted to be in the same direction as these hypothesized for the imagery variable described above.

Cognitive and Sensory Stimulation Seeking. Cognitive stimulation seeking is believed to be linked to semantic information processing, whereas sensory stimulation seeking is believed to be related to the processing of visual imagery (Ornstein 1977; Zuckerman 1979). Cognitive stimulation seekers were hypothesized to be more receptive to verbal stimuli, while sensory stimulation seekers were predicted to respond better to visual stimuli. Similarly, we speculated that "cognitive" subjects would rate verbal ads as more familiar, with the opposite effect displayed by "sensory" subjects.

Holistic/Projective Response and Analytical/Instrumental Response. Holistic/projective response implies self-projection into an external event; the consumer is a passive receptor of incoming information. Conversely, analytical/instrumental response consists of active, manipulative reasoning regarding incoming data. It was believed that persons preferring to respond in an holistic fashion would prefer visual advertisement formats; whereas those preferring analytical response would favor verbal advertisement formats.

Sample

Data were gathered by distributing structured questionnaires to graduate and undergraduate students enrolled in behavioral science classes. In partial fulfillment of course requirements, students were required to complete one questionnaire themselves and to have four other individuals, who were not students, complete a questionnaire as well. Students were given extensive instruction in questionnaire administration procedures, before engaging in data gathering. All subjects completed the questionnaire anonymously and were informed that the data would be used for academic research purposes. No remuneration was provided. Questionnaires were carefully checked for incorrect or falsified response patterns. In three instances, questionnaires were discarded for this reason.

In total, 162 subjects completed all the research instruments in a satisfactory manner. Sample mean age was 29; mean educational attainment was 14.9 years. Gender was approximately equally divided between men (46.3%) and women (53.72), thus balancing any sex-linked response tendencies which may have confounded prior research (cf. Appel, Weinstein, and Weinstein 1979; Rossiter 1980). Demographic variables exerted no systematic effects, and will not be further discussed.

ANALYSIS AND FINDINGS

Subjects were administered the set of control variables (e.g., imagery, emotional involvement) and then administered the experimental advertising stimuli and response instruments. Subjects were given either all verbal or all-visual advertising stimuli. It was felt wise not to mix the two types of stimuli given to a single subject, because of possible carry-over effects. Hence, subjects could be classified into exclusively all-verbal or all-visual treatment groups. Because of random assignment, there were no significant differences between the treatment groups in any of the control variables or demographic variables measured (age, occupation. education. sex). In total, 79 subjects were contained in the all-verbal treatment group versus 83 in the all-visual group.

The primary analytical technique was classical ANOVA using advertisement format (verbal versus visual) as a bi-level factor and each of the control variables, in turn, as a second bi-level factor. So that all data could be analyzed in a factorial design (as opposed to a regression analysis). control variables were split at the median to create dichotomous (high, low) conditions.

Semantic Differential Scales

Results for the three semantic differential criterion scales (utilitarian/rational, aesthetic/emotional, familiarity) are given in Table 2. This table shows the significance levels and directionality of the mean deviation (+,-) of the two advertising formats, V (visual) or W (verbal), with respect to a particular semantic differential criterion scale (e.g., utilitarian/rational). A + sign indicates that mean ratings on a criterion scale were higher for a particular format; a - sign implies that mean ratings w ere lower. In each row a different control variable (e.g.. imagery) was included in the analysis as a second factor. The amount of explained variance for each analysis is provided by the eta2 statistic. Significant interactions between the control variable and advertisement format are indicated by dark corners in the appropriate cells of the table.

Before discussing the specific hypotheses, two general observations should be made. First, in virtually no instances were any of the control variables significantly related to the semantic differential criterion scales. Thus, the possession of these traits and abilities does not seem to influence systematically subjects' response tendencies on the three criterion dimensions examined. This finding, albeit not anticipated, is encouraging, for it suggests that study results here and in other investigations may not be affected directly by these characteristics. As a caveat to this conclusion, it should be noted that the decision to employ median splits rather than ANCOVA may have reduced the power of the analysis to reveal significant influences. Nonetheless, the pervasive lack of effects strongly implies the control variables' inability to mediate processing.

A second general observation is that in 10 (out of 48) instances significant interactions (p<.12) were present between the control variables and advertisement format. Thus, in approximately 20 percent of the analyses advertisement format and a particular control variable were found to be reactive. In these ten instances, the relationship between advertisement format and the criterion scale is not interpretable as a direct relationship and should be viewed cautiously. However, even though some control variables did interact with advertisement format, these instances did not seem to be linked to aberrational main effects between format and the criteria examined. With these observations made, let us now turn to a discussion of the specific hypotheses.

H1: It was hypothesized that subjects viewing all-verbal advertisements would rate them as being more utilitarian/ rational than subjects viewing all-visual advertisements. As column A in Table 2 indicates, this hypothesis was consistently confirmed. The utilitarian/rational ratings for the all-verbal form of the two advertisements were consistently above the combined visual-verbal average (indicated by a + deviation); whereas the utilitarian/rational ratings for the all-visual form of the same advertisements were consistently below the combined visual-verbal means (indicated by a - deviation). This condition held regardless of moderator variable controls and was consistently significant at the p < 001 level or beyond. This provides support for the a priori reasoning that verbal stimuli would be perceived was more utilitarian/rational than visual stimuli. We attribute this perception to the presence of cultural norms which equate language with rationality (Polanyi and Prosch 1976). An alternative explanation may be that semantic data tends to activate the more "rational" left cerebral hemisphere, leading to a perception of heightened rationality/utility (Hansen 1981).

H2: The second hypothesis stated the expectation that subjects viewing all-visual advertisements would rate them higher on the aesthetic/emotional dimension than subjects viewing all-verbal advertisements. As can be seen from the data in-column B of Table 2, the results regarding this hypothesis are ambiguous. For the second advertisement (Stendig International dining table), the hypothesis was marginally confirmed (p< .10) in all but one instance. [The exception was when external sensory stimulation seeking served as a moderator.] However, the significance levels are not as high as was found for the first hypothesis. The results pertaining to the other stimulus advertisement (figure of the goddess Fortuna), however, were significantly counter to the hypothesis. In this instance, subjects who viewed the all-verbal version of the advertisement rated it higher in aesthetic/emotional content than subjects who viewed the all-visual version of this same advertisement.

Satisfactory explanations for this unexpected reversal are difficult to generate. For example, it is difficult to argue that the few emotion-laden words contained in the copy of the all-verbal version of this advertisement (e.g. "naked," "amidst the waves") are somehow more stimulating of aesthetic/emotional response than the actual image of the unclad woman in the all-visual version. Further the second advertisement, for which the expected effect was found, also contained affective copy (e.g. "most beautiful environments," "elegance incarnate," "Soaring,t' "dramatic"), and featured a round table topped with flowers in the all-visual version. Hence, we are unable to identify a sound explanation for the ambiguity of the findings regarding the second hypothesis.

H3: More encouraging results were obtained for the third hypothesis, which posited that subjects viewing all-visual advertisements would rate them as being more familiar than subjects viewing all-verbal advertisements. The data in column C of Table 2 clearly confirm this hypothesis at significance levels ranging from .02 to .001. Generally, the first advertisement (of goddess Fortuna) was rated higher on familiarity than the second (of the Stendig table); but in both cases deviations from the combined visual-verbal mean were positive for the all-visual treatment and negative for the all-verbal treatment.

These findings provide consistent support for an hypothesis that originated largely on a speculative basis--that is, because images are easier and more comfortable to process cognitively, people will believe that a novel visual stimulus is more familiar than a similar one of verbal content. In a sense, this belief is not developmentally inaccurate, for visual images are among our earliest sources of information about the world. Language, especially literacy, is a cultural artifact acquired at a much later ontogenetic stage and is, therefore, less familiar in both its recency of usage and artificiality of form.

CONCLUSIONS AND IMPLICATIONS

This study attempted to deal with some unresolved issues in past work on the effects of advertising format on consumer response. First, prior research has been hampered by the potential confounding of medium effects with message format effects. The present study utilized an all-verbal versus an all-visual format within one type of medium (i.e., print) to overcome this problem. Second, product type was held constant across both formats--a condition not met by some prior studies. Finally, actual advertisements were used, instead of researcher-created ones, in an effort to instill greater realism into the experiment.

The data were generally supportive of the overall prediction that the format of an advertisement (verbal versus visual) would mediate subjects' perceptions of the advertisement, even when the information conveyed by each format was generally consistent. These effects appear to be largely independent of the several moderating variables employed here in. Any conclusions must be tempered by the acknowledgement that some of the attained effects were stimulus-specific (i.e., the inexplicable reversal obtained for the aesthetic ratings of one ad). Overall, however, the replication of effects obtained across ads within the study is reassuring in this regard. Future research of this nature should explicitly consider the nature of the products being advertised, (e.g., degree of product involvement vis-a-vis the specific subject population; tangibility of the product or service being advertised) in addition to the format of the communication.

TABLE 1

INTERNAL CONSISTENCY OF CRITERION MEASURES

TABLE 2

UTILITARIAN, AESTHETIC, AND FAMILIARITY RESPONSE AS A FUNCTION OF ADVERTISEMENT FORMAT AND MEDIATING VARIABLES

REFERENCES

Appel, V., Weinstein, S., and Weinstein, C. (1979), "Brain Activity and Recall of TV Advertising," Journal of Advertising Research, 19, 4, August, 7-15.

Bower, Gordon H. (1970), "Imagery as a Relational Organizer in Associative Learning," Journal of Verbal Learning and Verbal Behavior, 9, 529-33.

Bowers, K.S. and P.G. Bowers (1972), "Hypnosis and Creativity: A Theoretical and Empirical Rapprochement," in E. Fromm and R.S. Shor (Eds.), Hypnosis: Research Development and Perspectives, Chicago: Aldine-Atherton.

Childers, Terry L. and Michael J. Houston (1983), "Imagery Paradigms for Consumer Research: Alternative Perspectives from Cognitive Psychology," Advances in Consumer Research, Vol. 10, (eds.) A. Tybout and R. Bagozzi, Ann Arbor. Association for Consumer Research, forthcoming.

Golden, Linda L. and Keren A. Johnson (1983), "The Impact o Sensory Preference and Thinking Versus Feeling Appeals on Advertising Effectiveness," Advances in Consumer Research, Vol. 10, (Eds.) T. Tybout and R. Bagozzi, Ann Arbor: Association for Consumer Research. forthcoming.

Hansen, Fleming (1981), "Hemispheral Lateralization: Implications for Understanding Consumer Behavior," Journal of Consumer Research, Vol. 8, 1, June, 23-36.

Holbrook, Morris B. and Robert B. Zirlin (1983), "Artistic Creation, Artworks and Aesthetic Appreciation," in R. Belk (Ed.), Nonprofit Marketing, Vol. 1, JAI Press, forthcoming.

Holbrook, Morris B. (1978), "Beyond Attitude Structure: Toward the Informational Determinants of Attitude," Journal of Marketing Research, 15 (November), 545-56.

Krugman, Herbert E. (1971', "Brain Wave Measures of Media Involvement," Journal of Advertising Research, 11, 1, February 3-9.

Lutz, Kathy A. and Richard J. Lutz (1977), "Effects of Interactive Imagery on Learning: Application to Advertising," Journal of Applied Psychology, 62, 493-98.

Lutz, Richard J., Scott B. .Mackenzie, and George E. Belch (1983), "Attitude Toward the Ad as a Mediator of Advertising Effectiveness: Determinants and Consequences," Advances in Consumer Research, Vol. 10, (Eds.) A. Tybout and R. Bagozzi, Ann Arbor: Association for Consumer Research, forthcoming.

Mitchell, Andrew A. and Olson, Jerry C. (1981), "Are Product Attribute Beliefs the Only Mediator of Advertising Effects on Brand Attitude?" Journal of Marketing Research, Vol. 18, August, 318-332.

Moore, Danny L., and J. Wesley Hutchinson (1983), "The Effects of Ad Affect on Advertising Effectiveness," in R.P. Bagozzi and Alice M. Tybout (eds.), Advances in Consumer Research, Vol. 10, Ann Arbor: Association for Consumer Research, forthcoming.

Osgood, C.E., Suci, G.J., and Tannenbaum, P.H. (1957), The Measurement of Meaning, Urbana: University of Illinois Press.

Balance of references available from authors.

----------------------------------------