Consumer Perceptions As a Function of Product Color, Price and Nutrition Labeling

Robert A. Peterson, The University of Texas at Austin
ABSTRACT - A 33 completely-crossed factorial experiment was conducted to assess the effect of color, price, and nutritional information on consumers' perceptions of bread. Subjects were 270 middle class housewives. Dependent variables were perceived nutritional value of bread and percentage of whole wheat contained in bread. The major finding was that color of bread was relatively more important in determining consumer perceptions than were price and nutritional information combined.
[ to cite ]:
Robert A. Peterson (1977) ,"Consumer Perceptions As a Function of Product Color, Price and Nutrition Labeling", in NA - Advances in Consumer Research Volume 04, eds. William D. Perreault, Jr., Atlanta, GA : Association for Consumer Research, Pages: 61-63.

Advances in Consumer Research Volume 4, 1977   Pages 61-63


Robert A. Peterson, The University of Texas at Austin

[Special appreciation is expressed to Liz Berezovytch, Marsha Richins and Beverly Spikes for their assistance in this research effort.]


A 33 completely-crossed factorial experiment was conducted to assess the effect of color, price, and nutritional information on consumers' perceptions of bread. Subjects were 270 middle class housewives. Dependent variables were perceived nutritional value of bread and percentage of whole wheat contained in bread. The major finding was that color of bread was relatively more important in determining consumer perceptions than were price and nutritional information combined.


It is an understatement to say that Americans are becoming increasingly concerned about the nutritional value of the food they consume. This concern has in turn manifested itself in numerous forms. There have been the vocal cries of consumer groups; there has been increased legislation; and certain consumers have even, to some extent, changed their eating habits. Moreover, there has been a virtual explosion of research devoted to specific nutrition-related issues.

One focus of this research has been the physical properties of food, its actual nutritional value in a consumption sense. A second focus has dealt with such topics as the perceived nutritional value or the amount of nutritional information necessary for consumers to make "satisfactory" purchase decisions.

The present research falls into the latter category. Specifically, this study was concerned with a commonly consumed and researched product--bread. Bread was selected as the study product for several reasons, including its widespread consumption and its being the current focus of much consumerism-related legislation.

In a broader context, this research can be considered an attempt to evaluate the effects of selected intrinsic (e.g., color) and extrinsic (e.g., price) product characteristics on consumer perceptions. Hence the present work is but a variant of that research stream concerned with product quality evaluations (e.g., Peterson and Jolibert, 1976; Szybillo and Jacoby, 1974).

The Real Test

Squeezability, long a standard household test for assessing bread freshness, is now recognized as being an imperfect indicator. Nor is whiteness regarded as a sure sign of wholesomeness anymore. Rather, the opposite is likely to hold--dark colored bread is now often thought to be nutritionally superior to white bread, and the firmer the loaf, the higher bread quality. This is because dark breads, especially wheat breads, contain more whole grain products, and whole grains have been demonstrated to be dietarily superior to bread not so constituted.

But color alone is an insufficient indicator of nutritional value. The Center for Science in the Public Interest (a Washington-based consumer group) alleges that some breads contain coloring (both natural and artificial) to make them appear darker and hence more nutritious than they actually are. Likewise, a recent Consumer Report (1976) study reported that, as a class, neither white nor dark bread was nutritionally superior in a relatively large sample of breads.

In addition, price is no certain criterion for selecting the most nutritious bread since it is often nothing more than a marketing tool. While dark bread is typically more expensive than white bread, the Consumer Report study found that, among the brands investigated, the least nutritious breads were frequently the most expensive, and vice versa.

Often the only objective nutrition information is that supplied by the package label. By law the label must contain the ingredients of the bread as well as the percentage of daily recommended allowances of protein and seven essential vitamins and minerals (Beloian, 1973). However, there are two potential problems relating to these labels. The labels may not incorporate all the necessary information, such as the amount of whole grain wheat or the degree of roughage present, for efficient decision making. Second, there is some question as to whether consumers even understand or use information provided by current labels. In a recent review article, Day (1976) has noted that the only apparent effect of nutrition labeling has been an increase in consumer confidence when purchasing such labeled products.


While the above paragraphs summarize, in a most condensed form, much of what is known concerning certain nutritionally-oriented bread issues, there still remains a gap in knowledge. The purpose of the present research was to empirically determine the simultaneous effect of three factors--bread color, bread price, and nutritional information--on selected consumer perceptions. With this in mind, the study can be viewed as somewhat complementary to the work of such investigators as Friedman (1972), Asan and Bucklin (1973), and Lenahan et. al (1973).


Independent Variables

The study employed a 33 completely-crossed factorial experimental design. This resulted in an experimental layout of 27 cells, each containing a particular treatment combination. The three independent variables (factors) employed in the study were color, price, and nutrition information. Each was measured at three levels representing, respectively, low, medium, and high values of the factor.

Three color levels--"white," "light-wheat," and "dark-wheat"--were employed. Great care was followed in obtaining these various levels. All bread was prepared and baked by an independent baker at the same time.

Loaves were then sliced, and slices individually wrapped and frozen. With the exception of coloring (wheat colors were achieved by using closely measured amounts of burned brown sugar), all bread consisted of identical ingredients. While bread color was not rated on a color scale, the two wheat color levels corresponded very closely to "light" and "dark" wheat bread available in the marketplace. Preliminary tests revealed consumers had no difficulty in distinguishing among the three color levels.

Price levels were set at 294, 594 and 894 per 1 lb. loaf. These prices represented the low, median, and high prices of a 1 lb. loaf of bread available in the market area at the time of the study.

To obtain nutrition information a sample of 27 bread labels was studied. This ultimately resulted in the construction of three levels of nutritional values. The levels represented low, median, and high amounts of nutritional ingredients. The ingredients reported were those eight standard ones required by law: protein, vitamins A and C, thiamine, riboflavin, niacin, calcium, and iron.. They were reported in standard percentage terms of daily requirements per slice. Within each level percentages reflected the lowest, median, or highest percentage requirement being met for each of the ingredients by the various brands of bread currently offered at the time of the study. [The average difference between contiguous nutrition levels was 4 percentage points.]

In brief, the manner in which the factor levels were operationalized insured a minimum of artificiality in mirroring the marketplace. Pretests uncovered no questioning of the appropriateness of any of the factor levels employed.

Dependent Variables

There were two dependent variables employed in the study. These were perceived nutritional value of the bread, and percentage of whole wheat contained in the bread. The former was measured on a 7-point scale ranging from "very little nutritional value" to "very much nutritional value." The latter was measured on a 9-point scale varying from "none" to "about 100%."

These scales were embedded in the questionnaire with five other rating scales to minimize their importance. The other five scales dealt with such bread attributes as attractiveness, taste, and quality. To a limited extent the percentage of whole wheat variable served an additional, functional purpose--a manipulation check.


Subjects were middle class housewives living in two geographic areas (four census tracts) in a southwestern urban area. Two hundred seventy subjects were employed in the investigation. Each subject was assigned to one of the 27 experimental cells on the basis of a systematic random sampling procedure such that there were 10 subjects per cell.

Data were collected by means of personal interviews conducted by seven interviewers (student research assistants) at the residence of the subject. Interviewing was conducted over a 17 day period during daylight hours only.

Upon agreeing to participate, each subject was shown an individually wrapped slice of bread on a white tray under the guise of a new product evaluation study. She was then handed a simulated label (5" x 8" card) containing the price of a 1 lb. loaf of bread as well as nutrition information. No brand or store information was provided. Subjects were not permitted to touch the bread. The reason for presenting a single slice rather than an entire loaf was that use of an entire loaf did not permit administration of the color factor without undue handling. Each bread slice was only used in a single treatment combination.

Subjects were then administered a brief questionnaire containing the dependent variable scales, demographic items, and two open-end questions concerning bread purchasing in general. After completing the questionnaire subjects were thanked for their participation and debriefed as to the purpose of the study.


Data were analyzed using standard analysis of variance (ANOVA) techniques with one exception. The observed error term ("O") was decomposed into a "corrected" error term ("C") and an "interviewer effect" term. This latter term was necessary to control for the effect of the seven different interviewers used in collecting the data. It can be viewed equivalent conceptually to a "blocking-covariate" factor.

Table 1 presents the results of analyzing the nutritional value scale. Only the color main effect was statistically significant (p < .03). While the color-nutritional value relationship was essentially monotonic (XW = 3.75; XLW= 4.15; XDW= 4.34), the only significant pairwise difference was between the white and dark-wheat colors.



Table 2 contains results from analyzing the second dependent variable. Again, the color main effect was statistically significant (p < .0001). In this instance the respective means were:

XW = 2.83

XLW = 4.08

XDW = 4.56

with the white color mean being significantly different from the light and dark-wheat means. It should be noted in passing that this result supports at least a somewhat effective treatment manipulation.



There was also a significant relationship between nutrition level and perceived amount of whole wheat contained in the bread (p < .01). Here, the only significant pairwise difference existed between the low and high levels of nutrition information.


While the results of this study are not broadly generalizable, there are several inferences which can be gleaned. First, few of the effects investigated were statistically significant. Conceptually this may be due to a lack of treatment impact. In other words, the treatment manipulations simply may not have been effective. This may be indicative of a poorly designed experiment or a lack of subject involvement in the study or even a lack of subject commitment in the marketplace. [As an alternative, a comparative (quasi-experimental) design allowing direct relative evaluations could have been employed to overcome any potential weakness of effect due to a monadic treatment presentation.]

Statistically, the lack of significance might be attributed to relatively large error (intra-cell) variation. Such a phenomenon is somewhat common in quasi-field experiments, like the present investigation, where there is a lack of strict control. Note that this may also relate to the lack of subject involvement in the research process, or large inter-interviewer variances.

Still, in spite of this general observation of rather meager findings, the present study has definite value. To a limited extent study results support those obtained by the Center for Science in the Public Interest; bread color had a relatively greater impact upon dependent variable responses than did price and nutrition information combined. Hence it may well be time, as the Center argues, to seriously consider legislation relating to the coloring of bread.

The results also have numerous implications in a wider sense. They support the findings of Szybillo and Jacoby (1974) and Peterson and Jolibert (1976) on the relative unimportance of price as a perceptual determinant in a multi-cue situation. And, they reinforce the unsettling notion that nutrition information is unimportant and/or not used by housewives in their purchase decisions.

Obviously, there still remains a need for research in this area; not only is there a need with regard to the particular issue dealt with here, but there is a need for issue-oriented research in general. Hopefully the present study can function, at least in some small measure, as a partial prototype; both in concept and in methodology. For instance the process followed here

experimental manipulation

relevant subject pool

multivariable approach

nonlaboratory context

has much to offer for an area oftentimes fraught with emotionalism and poor research methodology.

Finally, there are broad implications for consumer education from this study. Assume, for example, that color is in fact used as a surrogate for nutrition information. Then, given the growing concern with what one eats, attention should be directed toward persuading housewives to attend more to the label and less to color in making their (bread) purchases.


Edward H. Asam and Louis P. Bucklin, "Nutrition Labeling For Canned Goods: A Study of Consumer Response," Journal of Marketing, 37(April, 1973), 32-37.

Arletta Beloian, "Nutrition Labels: A Great Leap Forward," FDA Consumer (September, 1973), 1.

"Bread: You Can't Judge a Loaf by Its Color," Consumer Reports, 41(May, 1976), 256-260.

George S. Day, "Assessing the Effects of Information Disclosure Requirements," Journal of Marketing, 40(April, 1976), 42-52.

Monroe Peter Friedman, "Consumer Responses to Unit Pricing, Open Dating and Nutrient Labeling," Proceedings, Association for Consumer Research, M. Venkatesan, ed. (Chicago, 1972), 361-369.

R. J. Lenahan, J. A. Thomas, D. A. Taylor, D. L. Call and D. I. Padberg, "Consumer Reaction to Nutritional Labels on Food Products," Journal of Consumer Affairs, 7(Summer, 1973), 1-12.

Robert A. Peterson and Alain J. P. Jolibert, "A Cross-National Investigation of Price and Brand as Determinants of Perceived Product Quality," Journal of Applied Psychology, 61(August, 1976), 533-536.

George J. Szybillo and Jacob Jacoby, "Intrinsic Versus Extrinsic Cues as Determinants of Perceived Product Quality," Journal of Applied Psychology, 59(February, 1974), 74-78.