Effects of Picture Size and Color on Brand Attitude Responses in Print Advertising

Larry Percy, CREAMER INC
John R. Rossiter, N.S.W. Institute of Technology
ABSTRACT - Recent studies have suggested that visual elements in advertising, functioning independently of verbal elements, mediate belief, attitude, and intention responses toward the brand advertised. This paper reports a study that extends this work by examining the effects of two major visual variables in print advertisingCpicture size and color vs. black and white C on consumer attitudes toward a fictitious brand of mineral water. Results indicate there is a significant picture size effect upon multiattribute attitude toward the brand; and a significant color effect upon overall affect toward the brand. Implications of differing attitude models also. are discussed.
[ to cite ]:
Larry Percy and John R. Rossiter (1983) ,"Effects of Picture Size and Color on Brand Attitude Responses in Print Advertising", in NA - Advances in Consumer Research Volume 10, eds. Richard P. Bagozzi and Alice M. Tybout, Ann Abor, MI : Association for Consumer Research, Pages: 17-20.

Advances in Consumer Research Volume 10, 1983      Pages 17-20


Larry Percy, CREAMER INC

John R. Rossiter, N.S.W. Institute of Technology


Recent studies have suggested that visual elements in advertising, functioning independently of verbal elements, mediate belief, attitude, and intention responses toward the brand advertised. This paper reports a study that extends this work by examining the effects of two major visual variables in print advertisingCpicture size and color vs. black and white C on consumer attitudes toward a fictitious brand of mineral water. Results indicate there is a significant picture size effect upon multiattribute attitude toward the brand; and a significant color effect upon overall affect toward the brand. Implications of differing attitude models also. are discussed.


Although there has been evidence for some time that visual factors can affect the recognition and recall of advertising (c.f. Hendon 1973; Lutz and Lutz 1977; Ramond 19761, very little work has been done that addresses the question of whether visual factors influence not only responses to the ad itself but also responses to the advertised product, such as brand beliefs, attitudes, and intentions.

In a discussion of the effects of visual imagery responses to advertising, Rossiter and Percy (in press) have noted that a number of researchers such as Paivio (1971) have found strong support for the proposition that visual stimuLi almost inevitably elicit visual responses of the memory variety. In a well-known experiment by Shepard (1967) it was suggested that visual memory always occurs m response to visual stimuli. Subjects in his experiment studied 612 illustrated magazine advertisements at their leisure, and were given a recognition test on 68 pairs of ads in which the pairs consisted of one advertisement from the originally studied set of 612, plus a new ad of the same type. The task was to recognize the ads from the original set. When asked for immediate recognition, 98.5% of the subjects were accurate; one week Later, 90% were still accurate in their recognition. It required four months before recognition reached the chance level of 50%.

However, Paivio's and Shepard's findings that visual stimuli increase recognition and recall can not be taken as evidence that visual stimuli can also increase attitude, since attitude was never used as the dependent variable in any of those studies. Three recent studies, however, have used attitudinal dependent variables. These are summarized below. The first two studies examined isolated aspects of visual content in advertising. The third was somewhat more systematic and farms the basis for this present experiment.

A study by Mitchell and Olson (1977; 1981) looked, among of.her things, at the influence upon attitudinal responses of two types of hypothetical print advertisements; one with explicit verbal claims, and one visual with no verbal cue beyond a fictional brand name. They found that the visual ad (basically a picture or a kitten paired with a fictitious brand name for a bathroom tissue) was more effective than the ad with only an explicit verbal claim in communicating the target belief of "softness" for the brand. Additionally, the visually oriented ad was more effective in generating positive overall affect toward the brand. However, another visual ad for the tissue (a picture of a sunset) also significantly increased positive coverall affect without having any influence on beliefs. This finding is relevant and will be returned to in the discussion.

A second study, conducted by Wright (1979), went beyond print to television in investigating the effect of concrete visual and verbal messages upon actual behavior. Specifically, verbal and visual messages were constructed to influence consumers to read on-package warning instructions for over-the-counter drugs. The two visual variations presented a weak vs. strongly concrete execution: a focus upon the package vs. a consumer securing and actually reading the back of a package. Wright found that the combination of concrete verbal with the concrete visual enactment produced a significantly stronger effect upon behavioral compliance with the warning. This study combined both visual and verbal elements, centering upon the effects of using more concrete language and concrete visual situations to induce superior imaging.

The third study, by Rossiter and Percy (1978, 1980), adopted a more systematic approach to examining visual stimuli by varying the visual content of the advertising along a single stimulus dimension: picture size. They found that a large picture (of a brimming mug of a hypothetical new brand of beer) produced almost twice as favorable effect on overall brand attitude than did an identical but smaller picture. However, there were several limitations with this experiment: (1) the visual manipulation was experimentally confounded with a verbal manipulation involving Concrete copy claims, as in the Wright (1979) experiment, such that the interaction between visual and verbal factors could have been the true causal mechanism; 12) picture size was varied at only two levels, raising the issue of whether the picture size phenomenon is linear or follows some other form, such as the commonly reported "square root law" for ad size and memory; and (3) the only attitudinal dependent variable was overall affect toward the brandC specific beliefs about the brand were not measured.


The present experiment extends Rossiter and Percy's earlier study by: reexamining the picture size phenomenon for a different product; varying picture size through three levels (though the form-of-effects analysis is not reported here); varying only picture size in independent treatments, i.e., with constant copy; including brand beliefs as well as overall affect as dependent attitudinal measures; and systematically adding a second visual stimulus dimension, color (the earlier Rossiter and Percy experiment used only black and white whereas the Mitchell and Olson experiment used only color ads). Theoretical reasons for selecting size and color as the visual stimulus dimensions are respectively outlined next.

Picture size has long been recognized in psychology (Kosslyn 1980) and in advertising (Starch's "square root law") as a visual stimulus variable that increases memory responses such as recognition and recall. However, it is less obvious why picture size should also positively influence attitude responses as in the Rossiter and Percy (1978; 1980) experiment. We propose three related explanations as follows. Firstly, a larger picture of the product in an ad may cause consumers to focus on the product rather than the ad, thus generating a stranger product attitude than if a smaller picture of the product were shown. This might be called the "attention" explanation. Secondly, larger pictures produce larger reported visual images that tend to block out competing images (Kosslyn 1975); these may persist during the post_exposure rating task and additionally focus the rating on the product. This may be called the "image carryover" explanation. Thirdly, a larger picture of the product makes it appear more lifelike in size and may therefore more readily stimulate thoughts or images of consumption. This may be called the "consumption imagery" explanation. All three phenomena may, of course, operate together. The attention and image carryover explanations are essentially cognitive whereas the consumption imagery explanation is essentially affective. Accordingly, we might expect picture size, on balance, to influence general belief-based attitude more than affective attitude toward the Product.

Color was selected as an important and commonly used visual advertising stimulus dimension. Some years ago, two psychologists at Xerox Corporation (Dooley and Harkins 1970) demonstrated that color's principal effectiveness lies in transmitting emotion, whereas a black and white presentation is equally as effective in transmitting information. Thus we might expect color to have almost the opposite effect to picture size an the respective attitude responses. A color version of an ad .may positively influence affective attitudinal response but not necessarily belief-based responses to the product.


The product category selected for this experiment was mineral water. A fictitious name, Esprit, was selected and presented as a possible new imported brand of mineral water. The stimulus ads used in the study were based upon a professionally prepared "rough" execution, i.e., artwork rather than a finished graphic or photography. The advertisement itself consisted of the brand name, Esprit, at the top of the page, with a rendering of a bottle and glass beneath (with minimal background). Variations of this advertisement were then created by making color copies of the full picture, a 30% reduction of the picture (but not the brand name), and a 60% reduction of the picture.

This provided three stimulus ads: and full-size, one with a medium sized visual, and one with a small visual, all in full color. Black and white versions were created by simply reproducing the three ads with a conventional copier. (see Figure 1)


The study itself was conducted among 90 adult heads of households (45 men and 45 women) randomly selected for personal mall-intercept interviews at a typical suburban shopping mall in a midwestern city. A broad range of demographics was represented. Each subject was randomly assigned to one of 6 cells representing the three size and the two color vs. black and white conditions. The test itself was monadic, so that each subject saw only one ad. The full experimental design was thus a 3 x 2 factorial With 15 observations per cell. Statistical power considerations, and the exploratory nature of the study, suggested that the alpha level of significance be set at .10, although exact probabilities are reported.

Subjects were presented one of the advertisements and asked to look at it and read it as they would normally read an advertisement in a magazine. The ad was removed, and the subjects were asked a short series of elicitive probes in an attempt to gather imagery data and cognitive response data (not reported here). They were then asked to look it the ad again, and based upon it, to rate the extent to which they believed Esprit possessed five attributes. Subjects were handed a card with 11-point bipolar rating scales denoting the following attributes (brand beliefs): socially acceptable socially unacceptable; good tasting - poor tasting; expensive - cheap; natural - artificial; chic -"pseudo-chic." These attributes were selected from pilot research an the frequency of freely elicited attributes for the mineral water category. Additionally, subjects were asked to rate the personal importance of each of these attributes when they choose a mineral water product, using a 0 (Very Unimportant) to 10 (Very Important) scale. Next, a measure of overall affect toward Esprit, based on the advertisement, was gathered over a 7-point scale for four dimensions: good - bad; inferior superior; unpleasant - pleasant; interesting - boring. These four dimensions have been found in previous research (e.g., Fishbein and Ajzen 1975; Rossiter and Percy 1980) to yield a unidimensional measure of affect with reliability (coefficient alpha) of over .80. Finally, each subject was asked how likely it would be that they would try this brand of mineral water, using a scale of 0 (would not try) to 10 (would try).

In addition to considering the individuAl attribute beliefs, the five scales were summed for an overall measure of beliefs. Also, following the multiattribute attitude formulations, the summated product of the beliefs and their corresponding importance weights was computed. An overall affective measure was computed from the sum of the four evaluative measures.


The Anova main effects of the three size and two color variations on individual brand beliefs, summed beliefs, multiattribute attitudes affect, and intentions are summarized in Table 1.



They suggest very little effect an individual beliefs, although a tendency toward main effect for size is apparent when all five beliefs are summed (p = .134). While not as strong as the general finding of Mitchell and Olson (1977), there is at least evidence that such variations in visual elements as size can have an effect upon beliefs. It is also interesting (and at least consistent with common sense) to see that color vs. black and white can have a significant main effect upon consumer's taste beliefs. This would suggest that while color in advertising may not influence all beliefs, it could have an influence where relevant imagery is involved, as would be the case with taste.

The multiattribute attitude measure, as modeled by the sum of beliefs weighted by importance, was significantly influenced by size (p = .008). These results are further detailed in Table 2. It would seem that these findings support the earlier results of Rossiter and Percy (1978), which implied that a larger picture generated more favorable brand attitude than a smaller picture.



Interestingly, however, as shown in Table 3, the measure of overall affect toward the brand snowed no significant size main effect (p = .697), although it did show a significant color main effect (p = .0011, and a slight color by size interaction (p = 059). Failure to find a main effect for size is surprising, because the original attitude measure reported by Rossiter and Percy was in fact a measure of affect identical in all respects to the measure used here. This suggests that the verbal element contained m the advertisements used in their earlier study was indeed a confounding factor in the interpretation of visual effects following multiattribute attitude formulation



Finally, as shown in Table 1 (earlier), while there was a moderate main effect for color (p < .094) on intention to buy the new brand of mineral water, there was none for size (p < .325).


The results of this experiment both confirm and clarify the work reported by Mitchell and Olson (1977; 1981) and Rossiter and Percy (1978; 1980). It is clear that visual elements in advertising, without the aid of verbal components (beyond the brand name) can mediate belief and attitude responses to advertising. Further, attitude has been shown to be a function of the size of the visual element in the execution, where larger pictures of the product generate significantly more favorable attitudes than the same picture reduced in size. This confirms the picture size finding of Rossiter and Percy's earlier experiment without the confounding of verbal elements but in relation to a different measure of attitude (see below).

In addition, the use of color vs. black and white in print ads was found to significantly influence overall affect toward the brand, but not the consumer's belief-based multiattribute attitude. However, given the significant effect of color upon the taste belief, there is evidence that color can have specific effects on beliefs for which color cues in ads are logically relevant

Overall, he findings tend to support the idea that two commonly used visual elements in print advertising, size of picture and color of picture, may exert their influence in different ways. Picture size may produce a cognitive-perceptual type of processing whereby the product or its image in memory is given stronger focus with a larger picture, thus influencing primarily belief formation. Color, on the other hand, may produce a basically affective type of processing that influences emotional response to, but not beliefs about, the product (Mitchell and Olson 1981; Shimp 1981). More detailed investigation of the causal processes size and color effects is indicated.

A crucial question raised by these results is: which measure of "attitude" is correct? As discussed by Percy and Rossiter (1980) the answer depends entirely on the advertiser's assessment of "what" attitude-type elements in the consumer's head actually cause the consumer to try the brand. For many products and many consumers (e.g., aspirin for most aspirin users), one or two specific, important brand beliefs surely drive choice, not whether consumers affectively "like" the brand; for others, a relatively numerous composite of "rational" beliefs in the multiattribute sense may be causal; whereas for still other products, or other consumers, the brand's "likability" in a- purely affective sense may be sufficient to motivate brand choice. In our view, there should be less concern about which attitude model is "right" in general C instead, more attention should be paid to assessing the attitude model that is causally most appropriate to the product and target audience in question.

In the present study, picture size was found to be relevant if the target audience chooses the product an the basis of multiattribute attitude, whereas color was found to be more relevant if the choice is based on overall affect or perhaps a primary expectation (belief) about taste. Probably the best research procedure, given qualified effects of this nature emerging so frequently in attitude research, is to include multiple measure of attitude and to report the effects in full.

In summary, this extension of the initial work in the area of visual effects upon brand-relevant responses to advertising underscores the importance of the visual element, independent of verbal contributions, in producing attitude formation toward an advertised brand. This suggests that care must be taken in any test of the influence of advertising upon brand attitudes to adequately consider the influence of the visual elements in addition to the traditional measures associated with message copy. The problem, or the solution, may be in the picture.


Dooley, R.P. and Harkins, L.E. (1970), "Functional and Attention-Getting Effects of Color on Graphic Communications," Perceptual and Motor Skills, 31, 851-854.

Fishbein, M. and Ajzen, I. (1975), Belief, Attitude, Intention and Behavior: An Introduction to Theory and Research, Reading, MA: Addison-Wesley.

Hendon, D.W. (1973), "How Mechanical Factors Affect Ad Perception," Journal of Advertising Research, 13:4, 39-45.

Kosslyn, S.M. (1973), "Information Representation in Visual Images," Cognitive Psychology, 7, 341-470.

Kosslyn, S.M. (1980), Images and Mind, Cambridge, MA: Harvard University Press.

Lutz, K.A. and Lutz, R.J. (1977), "The Effects of active Imagery an Learning: Application to Advertising," Journal of ApPlied Psychology, 62:3, 493-498.

Mitchell, A.A. and Olson, J.C. (1977), "Cognitive Effects of Advertising Repetition," in Perrault, W.D., Jr., (;d.), Advances in Consumer Research: Vol. IV, Atlanta, Georgia; Association for Consumer Research, 213-220.

Mitchell, A.A. (1981), "Are Product Attribute Beliefs the Only Mediator of Advertising Effects an Brand Attitude?," Journal of Marketing Research, 18:3, 318-332.

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

Percy, L. and Rossiter, J.R. (1980), Advertising Strategy: A Communication Theory Approach, New York: Praeger.

Ramond, C. (1976), Advertising Research: The State of ,he Art, New York: Association of National Advertiser.

Rossiter, J.R. and Percy, L. (1978), "Visual Imaging Ability as a Mediator of Advertising Response," in Hunt, H.K. (Ed.), Advances in Consumer Research: Vol. V, Ann Arbor, Michigan: Association for Consumer Research, 621-629.

Rossiter, J.R. (1980), "Attitude Change Through Visual Imagery Advertising," Journal of Advertising, 9:2, 10-16.

Rossiter, J.R. (in press), "Visual Communication in Advertising," in Harris, R.J. (Ed.), Information Processing Research in Advertising, Hillsdale, New Jersey: Lawrence Erlbaum and Associates.

Shephard, R.N. (1978), "The Mental Image," American Psychologist, 33:2, 125-137.

Shimp, T.A. (1981), "Attitude Toward the Ad as a Mediator of Consumer Choice Behavior," Journal of Advertising, 10 9-15.

Wright, P. (1970), "Concrete Action Plans in TV Messages to Increase Reading of Drug warnings," Journal of Consumer Research, 6:3, 256-269.