Judgments of Verbal Versus Pictorial Presentations of a Product With Functional and Aesthetic Features

Teresa J. Domzal, George Mason University
Lynette S. Unger, Miami University
ABSTRACT - This study follows the methodology employed in Holbrook and Moore's (1981) study of feature interactions in consumer judgments of verbal versus pictorial presentations. It uses a more functional product, wrist watches, as the stimulus. In contrast to the findings in the original study, pictorial presentation of stimuli generated significantly fewer main effects and significantly fewer interaction effects in judgments of the stimuli than did verbal presentation. Individual cognitive strategies did not appear to explain these results as they did in the original study. Verbal loop, imagery coding and dual coding theories were utilized to explain findings.
[ to cite ]:
Teresa J. Domzal and Lynette S. Unger (1985) ,"Judgments of Verbal Versus Pictorial Presentations of a Product With Functional and Aesthetic Features", in NA - Advances in Consumer Research Volume 12, eds. Elizabeth C. Hirschman and Moris B. Holbrook, Provo, UT : Association for Consumer Research, Pages: 268-272.

Advances in Consumer Research Volume 12, 1985      Pages 268-272

JUDGMENTS OF VERBAL VERSUS PICTORIAL PRESENTATIONS OF A PRODUCT WITH FUNCTIONAL AND AESTHETIC FEATURES

Teresa J. Domzal, George Mason University

Lynette S. Unger, Miami University

ABSTRACT -

This study follows the methodology employed in Holbrook and Moore's (1981) study of feature interactions in consumer judgments of verbal versus pictorial presentations. It uses a more functional product, wrist watches, as the stimulus. In contrast to the findings in the original study, pictorial presentation of stimuli generated significantly fewer main effects and significantly fewer interaction effects in judgments of the stimuli than did verbal presentation. Individual cognitive strategies did not appear to explain these results as they did in the original study. Verbal loop, imagery coding and dual coding theories were utilized to explain findings.

INTRODUCTION

One objective of advertising is to communicate information to consumers that will result in positive evaluation of the advertisement itself and ultimately, the product. Major paths of theory development dealing with 'successful" advertising, typically in terms of recall and effectiveness, can be traced to research in the areas of attitude formation and change and information processing.

The area of information processing in advertising draws upon neurophysiological research such as brain wave measurement and left-brain, right-brain activity. This represents the most basic beginnings in understanding how people think, perceive and remember. Psychological theories of information processing focus on incoming stimuli (e.g., pictures, words, etc.), and how they are perceived, encoded, stored and remembered.

The present study is designed to ascertain differences in consumer information processing when subjects are presented with either visual or verbal stimuli and are asked to evaluate eight typical wrist watch designs.

BACKGROUND

Major theories tested in visual and verbal information processing overlap and to a certain extent, contradict each other. Information processing theory has roots in physiological functions of the human brain and in psychological theories that attempt to understand cognitive processes (Appel, Weinstein and Weinstein 1979; Bugelski 1977; Geschwind 1979; Hansen 1981; Paivio 1971; Paivio 1978; Richardson 1977; Rossiter and Percy 1980; Shephard 1967; Weinstein, Weinstein and Appel 1980).

Holbrook and Moore (1981) applied visual/verbal information processing theory in predicting assessments of an aesthetic product, sweaters, (also see Holbrook and Moore for a review of left-brain right-brain activity). They theorized that evaluation of such a product would be a "gestalt like phenomenon in which features combine configurally as well as additively." (p. 103). They provided support for the presence of cue configurality, or interaction effects in assessments of such products. They further presented 32 sweaters in either verbal or visual treatments and suggested that these treatments tend to be processed by discrete processing systems. Pictorial presentations are processed holistically by the right-brain hemisphere. They are judged simultaneously as gestalts, and should generate many feature effects and interactions. Conversely, verbal presentations are processed atomistically and sequentially by the left-brain and should generate fewer feature effects and interactions. The results of the Holbrook and Moore study generally support these hypotheses.

The current study replicates the methodology and analyses employed by Holbrook and Moore but applies them to a relatively more functional product, wrist watches.

METHODOLOGY

Wrist watches were chosen as the stimuli in this replication study since they could be judged on function as well as style. The sweaters used in the Holbrook and Moore study were judged only on design features, so respondents were making primarily aesthetic judgments. It might be argued, however, that the potency factor in the original study constitutes a functional dimension, since the higher loading items (light/heavy, smooth/rough, soft/hard) could be interpreted as describing the warmth a sweater provides in addition to tactile attributes. In contrast, the replication subjects assessed the watches on both aesthetic and functional criteria.

The watches were designed to differ on three features, each with two levels: leather versus metal band; round versus square face; digital versus analog function. These three features combined to yield eight different watches which were presented as the stimuli in either pictorial or verbal treatments. In contrast, the sweaters in the Holbrook and Moore study combined five features, each with two levels, for a total of 32 sweaters. Fewer features were used in this study to simplify the task of completing the questionnaire. Moreover, in explaining their findings, Holbrook and Moore posited that their subjects may have perceived the task of evaluating the 32 sweaters as difficult and consequently resorted to simplification strategies (attempting to be consistent or focusing on one or two main features across sweaters). Such simplifications would tend to blur the difference between holistic and atomistic processing. Reducing the number of stimuli to be judged was an attempt to prevent this problem.

Following the task design used by Holbrook and Moore, subjects were given either the eight verbal descriptions (words treatment) or the eight pictorial representations (pictures treatment) in a written questionnaire. Examples of the watch designs and verbal descriptions are shown in Figure 1. Respondents were asked to judge each watch on 28 seven-point semantic differential scales. Twenty pairs of bipolar adjectives representing the evaluation - activity - potency factor structure of Osgood, Suci and Tannenbaum (1957) were used. Additionally, eight items used specifically to describe watches were added, based on preliminary interviews with 12 graduate students. These final 28 items are shown in Table 1. While the instrument was not completely randomized, four different watch and bipolar item orders were presented, and no significant response bias on the bipolar items was indicated. Instructions to respondents were identical to those used in the original study. Subjects further answered questions concerning information processing strategies and product usage and provided demographic and psychographic data.

FIGURE 1

STIMULUS EXAMPLES

TABLE 1

GENERAL FACTOR STRUCTURE

A more heterogeneous sample was used in this replication, with the purpose of extending the generalizability of the findings. A convenience sample consisting of 110 respondents ranging in ages from 18 to 49 was used. Surveys were completed in one sitting and were administered by trained marketing graduate students. Respondents were assigned at random to the words treatment (n = 51) or the pictures treatment group (n = 59).

HYPOTHESES

The Holbrook and Moore study initially tested three hypotheses with respect to differential judgments between verbal and pictorial stimulus presentations:

H1: Highly subjective fashion related objects should tend to be processed as gestalts.

H2: The presentation of stimuli in pictures (words) should encourage the right (left) hemisphere's tendency toward holistic (atomistic) and integrative (analytic) processing, thereby producing a more global (restricted) focus on various design features.

H3: Pictures (words) should tend to engage the imagery (verbal) system to be processed simultaneously (sequentially) in the right (left) brain, and therefore, to produce relatively more (fewer) interaction effects.

In testing the first hypothesis, Holbrook and Moore predicted a tendency toward feature interactions in judgments of the 32 sweaters in both verbal and pictorial stimulus presentations, since fashion items would tend to engage the right hemisphere. In the replication study, a similar tendency toward cue configurality across both treatments was predicted, since watches are a fashion item. However, their more utilitarian nature would indicate that this tendency toward feature interactions might be weaker than with the sweaters used in the original study. As Holbrook and Moore discussed, where product benefits are primarily functional, the decision maker would be more likely to use an additive decision model. Watches lie somewhere between highly aesthetic and highly functional products. While no effort was made to compare watches and sweaters along such an aesthetic - functional dimension, the emergence of a utility dimension in the watch study factor analysis indirectly confirms the functional nature of the product.

Following Holbrook and Moore, Hypothesis 2 would be supported by a great number of significant main feature effects in response to the pictorial versus verbal treatment. Similarly, the third hypothesis would predict that the pictures treatment would generate more feature interactions in resPonses than words.

In the original study, subjects were recontacted and a second set of data was collected regarding task-related information processing strategy, in an effort to clarify results of the original hypotheses testing. In this study, subjects indicated agreement on a 7-point scale on four items. corresponding to the hypotheses below:

1. When I rated the watches, I made an effort to be very consistent in my answers.

2. When I rated the watches, I made an effort to simplify the task by focusing on one or two features.

3. When I rated the watches, I tried to add up their relative pluses and minuses.

4. When I rated the watches, I tried to form a mental picture of what each watch would look like.

Normalized scores on these items were used to test the following four hypotheses:

H4A: An emphasis on consistency should reduce the relative advantage of pictorial versus verbal presentation in encouraging feature interactions.

H4B: A strategy of simplification should also produce a smaller relative tendency for picture treatments to evoke feature interactions.

H4C: A tendency to try to treat the (product) features additively should decrease the relative tendency of pictures, as compared with words, to produce feature interactions.

H4D: Attempting to form mental pictures of the stimuli should also reduce the relative advantage of pictorial versus verbal displays in encouraging feature interactions.

In the present study, these data were collected in the original administration, which would presumably enhance the reliability of the results, and the same four hypotheses were tested.

ANALYSIS

This replication study modified the original data reduction methodology in order to maintain independence among the eight watches presented. Holbrook and Moore used principal components analysis with varimax rotation to reduce their 20 scales for each of the two treatment groups (words and pictures). The responses on the 32 sweaters were combined in each of these analyses. Theoretically, this procedure is justified, since the 32 sweaters probably adequately sample the domain of possible styles. However, since each subject's responses on each of the 20 bipolar scales are brought in 32 times, independence among stimuli is not maintained. Attempting to remedy this problem, sixteen different factor analyses were initially conducted in this study. Principal components analysis with varimax rotation was used to reduce the 28 bipolar scales for each of the eight watches within each words or pictures treatment.

Following Holbrook and Moore, pooled factor scores then served as dependent variables in individual level regression analytic conjoint analyses. The independent variables were the three treatments (leather/metal band; round/square face; digital/analog function) operationalized as dummy variables and the three two-way interactions of these variables. The complete model is described in pp. 107-108 of the original study. These analyses determined the number of significant main and interaction effects on each factor for each subject in each treatment group (4 factors x 110 subjects = 440 regressions). As in the original study, the number of significant main and interaction effects in the 440 individual regression analyses were aggregated across subjects for hypotheses testing.

Four regression analyses were used to test Hypotheses 4A-D. The dependent variable was the number of significant feature interactions across the four judgmental factors for each subject. The independent variables were treatment group (operationalized as a zero-one dummy variable indicating membership in the pictures or words treatment group), the subject's normalized rating on the task variable of interest (maintaining consistency, attempting to simplify, adding up features or forming mental pictures) and a treatment x task interaction term, whose coefficient was of primary interest in the testing of Hypotheses 4A-D. A complete model is given in Holbrook and Moore (p. 110).

RESULTS

Following Holbrook and Moore, the relative strengths of the main and interaction effects were judged by calculating the ratio of the Hays' omega-squared statistic for the strongest main effect divided by that for the strongest interaction effect for each subject on each judgmental factor. The median ratio in the original Holbrook and Moore study was 1. 06. They interpreted this as an indication of cue configurality in judgments, since the strongest interactions were about as important as the strongest main effects in assessing sweaters. In the replication, however, the median ratio was higher at l.89. This would suggest that the strongest main effects were nearly twice as important as the strongest interaction effects.This was predicted, considering the less aesthetic and more functional nature of watches.

As shown in Table 2, Hypothesis 2 was not supported. The pictures treatment showed fewer main effects than words (1.32 versus 2.00, p = .03). This contrasts sharply with the Holbrook and Moore results, which supported the second hypothesis at p = .05.

TABLE 2

MEAN SIGNIFICANT MAIN, INTERACTION AND COMBINED EFFECTS IN INDIVIDUAL-LEVEL REGRESSION ANALYSES OF FACTOR SCORES

Similar to the results of the original study, Hypothesis 3 was not supported. Pictures showed fewer interactions than words (.88 versus 1.53, p = .01). Holbrook and Moore had similar results directionally, but they were not statistically significant.

In testing Hypotheses 4A through 4D, results of the replication contrast strongly with those of Holbrook and Moore. Their results suggested that a cognitive strategy explanation was more plausible than a task simplification explanation, since Hypotheses 4C and 4D were most strongly supported. In the current study, the task simplification explanation received more support, since only Hypothesis 4B could be supported in the predicted direction (p<.10).

In testing Hypothesis 4A, the main treatment effect was significant (p<.01) and the main task effect (maintaining consistency) was not. Neither effect was significant in the original study. The interaction of treatment and task effect was significant (p<.10) as it was in the original study, but in the opposite direction of the hypothesis. Efforts to be consistent in judgments were found to 'promote the relative advantage of pictures over words in eliciting feature interactions (Holbrook and Moore p. 110) in both studies.

In testing Hypothesis 4B, Holbrook and Moore predicted that efforts to simplify the task would decrease the impact of pictures in evoking feature interactions, however, their results were significant (p<. 10) in the opposite direction. In the replication, by contrast, the interaction was significant in the direction predicted by Holbrook and Moore. The main effect for treatment group was not significant in the original study but was significant (p<.01) in the replication. Conversely, the main effect for task (attempting to simplify) was negative and significant in the original study and negative but not significant in the replication.

While Holbrook and Moore found strong support for Hypotheses 4C and 4D, neither were supported in the replication study. The interactions of treatment and task (adding up features in 4C; forming mental pictures in 4D) were not significant. The main effect for treatment was significant (p<.01) in both equations and the main effect for task was not significant in either.

DISCUSSION

Results of the replication study vary significantly from those in the original study on the testing of Hypotheses 2 and 3. Explanations for these differences may lie in methodological differences between this and the Holbrook and Moore study was well as in theory. First, this study uses a more functional product while the original research focuses on a highly aesthetic product. Second, the replication manipulates three attributes as opposed to five in the original study. Finally, a broader, less educated sample is used in the replication. The discussion below incorporates both methodological and theoretical explanations of the findings.

For Hypothesis 2, Holbrook and Moore anticipated and their results substantiated that pictorial presentations would generate more main feature effects than verbal presentations. Results in this study indicated the opposite: verbal presentations elicited more main feature effects in consumer judgments of watches than pictorial presentations. A possible explanation for this finding lies in Paivio's (1978) dual-coding hypothesis. Paivio suggested that there are two distinct symbolic systems, one specialized for processing pictorial information and the other for processing verbal information. He proposed that these two systems are interconnected and processing in one system can activate processing in the other. Rossiter and Percy 's (1980) dual loop theory suggests that both visual and verbal inputs could be processed through either a verbal belief process or a visual imagery process. Further theoretical support is provided by Bugelski's (1977) imagery coding theory. This contrasts with Holbrook and Moore's original premise that pictures would be processed through the right-brain, holistically as gestalts and words would be processed through the left-brain, atomistically and sequentially. It is important to note that Holbrook and Moore recognized that this approach was oversimplified but served as a starting point for research.

The replication results from the testing of Hypothesis 4D suggest that some type of dual loop phenomenon took place. The responses to the item When I rated the watches, I tried to form a mental picture of what each watch would look like, were skewed very high, relative to the responses on the other three information processing items. Regardless of treatment (pictures or words), some 66: of respondents marked a 6 or 7 on the 7-point scale. This would indicate that subjects from both treatments tended to form a mental picture of the watches they assessed. In cognitive terms, both pictorial and verbal inputs were being processed through visual imagery. This would explain why Hypothesis 4D was not supported and it would also explain why the words presentation would generate at least as many significant main feature effects as the pictorial presentation.

It appears that the functional nature of the watches product may have triggered the visual processing of the verbal presentation. Percy (1981) distinguished between concrete and abstract words:

Concrete words are generally described as those which refer to objects, persons, places or things that can be seen, heard, felt, smelled or tasted. Abstract words refer to those things that cannot be experienced by our senses...Concrete words tend to excite more visual imagery... (p. 108).

Because they are evaluated more in terms of functional attributes, watchesare moreconcrete. There was no need to experience them aesthetically, and from the pictorial and verbal presentations, subjects had all the information they needed to evaluate them. In contrast, the sweaters in the original study constituted a more abstract stimulus. Information vital to making aesthetic judgments, such as experiencing actual tactile or color characteristics was missing in both the pictorial and verbal descriptions. In another study using real sweaters, Holbrook (1983) noted that weaknesses in presenting such artificial representations of real products. Net, because of their more concrete nature, the watches tended to elicit more visual imagery regardless of the presentation treatment.

The results of Hypothesis 3 also contrasted with the original study. Holbrook and Moore suggested that pictorial presentations would generate more feature interactions in judgments than would verbal stimuli, and this was not supported. In the watch study, the words treatment was found to generate more interactions than the pictorial treatment. This finding might be explained the same way as the Hypothesis 2 results. Because of their functional rather than aesthetic nature, watches tended to be processed through a visual loop regardless of the presentation treatment. This would indicate right-brain, gestalt-type processing which would elicit feature interactions.

While this explanation adequately explains why verbal presentations might be expected to generate as many feature interactions and main effects as pictorial presentations, it fails to clarify why the verbal presentations elicit significantly more than the pictorial presentations. The results of Hypothesis 4B testing may shed some light. Supporting Holbrook and Moore's original hypothesis, efforts at task simplification (specifically, focusing on one or two features) was found to decrease the impact of pictures in evoking feature interactions. Holbrook and Moore's findings were significant in the opposite direction, suggesting again a difference inherent in the different types of products judged. Another explanation for this difference in the two studies may lie in sample differences. The replication sample contained a broader cross-section of people who were less-educated than the graduate students used in the original study. It may be that the average person might resort to task simplification in an effort to complete a lengthy questionnaire more readily than a graduate student.

Further research in assessing differences in consumer processing of verbal and pictorial information is needed. The present study indicates that some differences may be explained by the nature of the product (functional versus aesthetic) being evaluated.

REFERENCES

Appel, Valentine, Sidney Weinstein, and Curt Weinstein (1979), "Brain Activity and Recall of TV Advertising," Journal of Advertising Research, 19(August ), 7-15.

Bugelski, B.P. (1977), 'Imagery and Verbal Behavior,' Journal ut Mental Imagery, 1, 39-52.

Edell, Julie A. and Richard Staelin (1983), "The Information Processing of Pictures in Print Advertisements,' Journal of Consumer Research, 10 (June), 45-61.

Geschwind, Norman (1979), 'Specializations of the Human Brain,' Scientific American, 241 (3), 180-99.

Hansen, Flemming (1981), "Hemispheral Lateralization Implications for Understanding Consumer Behavior," Journal of Consumer Research, 8(1), 23-37.

Holbrook, Morris B., and William L. Moore (1981), "Feature Interactions in Consumer Judgments of Verbal Versus Pictorial Presentations," Journal of Consumer Research, 8(June), 103-113.

Holbrook, Morris B. (1983), "On the Importance of Using Real Products in Research on Merchandising Strategy," Journal of Retailing, 59(Spring), 14-20.

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

Paivio, Allan (1971), Imagery and Verbal Processes, New York: Holt, Rinehart and Winston.

Paivio, Allan (1978), "A Dual Coding Approach to Perception and Cognition," in Modes of Perceiving and Processing Information, eds. Herbert L. Pick and Elliot Saltzman, Hillsdale, N.J.: Lawrence Erlbaum.

Paivio, A., T.B. Rogers and P.C. Smythe (1968), "Why are Pictures Easier to Recall than Words?' Psychonomic Science, 11, 137-138.

Percy, Larry (1981), "Psycholinguistic Guidelines for Advertising Copy," in Advances in Consumer Research, Vol. 9, Andrew Mitchell (ed 1, Ann Arbor, Michigan: Association for Consumer Research, 107-110.

Richardson, Alan (1977), "Verbalizer-Visualizer: A Cognitive Style Dimension," Journal of Mental Imagery, 1, 109-26.

Rossiter, John R. and Larry Percy (1980), "Attitude Change Through Visual Imagery in Advertising," Journal of Advertising, 9(2), 10-16.

Shepard, R.N. (1967), "Recognition Memory for Words, Sentences and Pictures," Journal of Verbal Learning and Verbal Behavior, 6, 156-163.

Weinstein, Sidney, Curt Weinstein, and Valentine Appel (1980), "Brain Activity Responses to Magazine and Television Advertising," Journal of Advertising Research, 20( June ), 57-63.

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