A Comparison of the Usage of Numerical Versus Verbal Nutrition Information By Consumers

ABSTRACT - The objective of this paper is to compare verbal versus numerical presentations in facilitating the usage of nutrition information by consumers. Verbal nutrition information, due to its descriptive nature, is argued to be used to a greater degree than numerical nutrition information, which requires a reference point in order to be interpreted. Hypotheses are developed and tested to evaluate this proposition. The results of an experiment suggest that verbal information may have several advantages in terms of weight given to brand information in making judgments of healthiness, accuracy of subsequent ratings of brands based on attribute information, and accuracy of recall of brand information. These advantages appear to persist even when numerical information is presented with summary information to facilitate its interpretation. This research demonstrates several advantages for verbal when compared to numerical nutrition information. Implications for consumer research on nutrition information in particular and product information in general are discussed.

Citation:

Madhubalan Viswanathan (1996) ,"A Comparison of the Usage of Numerical Versus Verbal Nutrition Information By Consumers", in NA - Advances in Consumer Research Volume 23, eds. Kim P. Corfman and John G. Lynch Jr., Provo, UT : Association for Consumer Research, Pages: 272-281.

Advances in Consumer Research Voume 23, 1996      Pages 272-281

A COMPARISON OF THE USAGE OF NUMERICAL VERSUS VERBAL NUTRITION INFORMATION BY CONSUMERS

Madhubalan Viswanathan, University of Illinois

ABSTRACT -

The objective of this paper is to compare verbal versus numerical presentations in facilitating the usage of nutrition information by consumers. Verbal nutrition information, due to its descriptive nature, is argued to be used to a greater degree than numerical nutrition information, which requires a reference point in order to be interpreted. Hypotheses are developed and tested to evaluate this proposition. The results of an experiment suggest that verbal information may have several advantages in terms of weight given to brand information in making judgments of healthiness, accuracy of subsequent ratings of brands based on attribute information, and accuracy of recall of brand information. These advantages appear to persist even when numerical information is presented with summary information to facilitate its interpretation. This research demonstrates several advantages for verbal when compared to numerical nutrition information. Implications for consumer research on nutrition information in particular and product information in general are discussed.

Understanding how consumers use nutrition information is an important area in consumer research. Consumer researchers have examined how consumers use nutrition information presented on packages. In light of the new labeling requirements, several formats for presenting nutrition information have been tested including verbal presentations of nutrition information, and the presentation of nutrition information per serving of a brand along with the Daily Value (cf., Levy et al. 1991). The objective of this paper is to compare the usage of numerical versus verbal nutrition information. Hypotheses based on the proposition that verbal information is likely to be used to a greater extent than numerical information were tested in an experiment. The rest of the paper is organized as follows. Past research of relevance is briefly reviewed in the next section followed by a description of the hypotheses. The details of an experiment conducted to test the hypotheses are then presented followed by a discussion of the implications of this research.

REVIEW OF PAST RESEARCH AND HYPOTHESES

Review of Past Research

Numerical versus verbal presentations of information are of importance in the area of nutrition information as suggested by past research on these two types of information (cf., Scammon 1977; Levy et al. 1991). Past research on nutrition information has focused on the use of different formats, different types of reference information (i.e., U.S. RDA), or preprocessed information such as verbal information in facilitating the interpretation of nutrition information. Researchers have studied the use of different nutritional programs and different presentation formats in order to simplify processing of nutrition information by consumers (cf., Russo et al. 1986; Muller 1985; Levy et al. 1985). Some past research has focused on the effect of reference information and summary information. Moorman (1990) showed that the presentation of nutrition information with reference information in the form of percent of U.S. RDA increased ability to process and accuracy of comprehension. Levy et al. (1991) compared different formats for presenting Daily Reference Values. Viswanathan (1994) showed that summary information such as an average or a range can facilitate the interpretation of numerical nutrition information.

Research in consumer behavior on the processing and use of numerical and verbal product information has direct bearing on research on nutrition information. Venkatesan et al. (1986) suggested that numerical information derives its meaning in comparison with other numerical information and does not have any meaning by itself. On the other hand, verbal information has been argued to be more descriptive in nature (cf., Scammon 1977; Huber 1980; Viswanathan and Childers 1992). Huber (1980) argued and found that evaluations were made more frequently with verbal when compared to numerical information, due to the evaluative nature of verbal information. Viswanathan and Childers (1992) argued that verbal information is more descriptive than numerical information in that it conveys the relative location of a brand on an attribute in terms of highness or lowness.

Past consumer research suggests that some comparison process has to occur in order to interpret numerical information and understand its meaning whereas verbal information can be interpreted directly because of its descriptive nature. By directly conveying the relative location of a brand along an attribute, verbal nutrition information may be easier to use than numerical nutrition information. Therefore, in a setting where nutrition information is presented verbally on some attributes and numerically on other attributes, consumers may tend to use verbal information to a greater degree than numerical information because of the ease of using such information.

Some past consumer research has examined numerical and verbal presentations of nutrition information. Scammon (1977) compared nutrition information presented as verbal adjectives versus percent of U.S. RDA (e.g., "good" versus "35" percent of U.S.RDA on, say, the attribute protein content). The author found that the most nutritious brand was identified more accurately with verbal when compared to percentage information. The author argued that verbal information is relatively preprocessed due to its evaluative nature. Therefore it requires less processing when compared to percentage information which is relatively unprocessed. Viswanathan (1994), in a study that showed that summary information such as an average or a range can facilitate the interpretation of numerical nutrition information, also examined verbal presentations of information. Verbal versus numerical presentations with or without summary information were manipulated between groups of subjects. Verbal information appeared to have several advantages when compared to numerical information without summary information in terms of weight given to information in healthiness judgments, recall and recognition of brand information, and time spent on information. Such a pattern was not found for differences between verbal information, and numerical information with summary information.

The objective of this study is to directly compare the usage of numerical versus verbal nutrition information when information on some attributes of a brand is presented verbally and information on other attributes is presented numerically. Such a manipulation of numerical versus verbal presentations within subjects offers a way of directly comparing whether consumers use one form of information to a greater degree than the other. Direct comparison of the relative usage of verbal versus numerical information within subjects has important implications for consumer research. Situations arise commonly where nutrition information is available in numerical form on some attributes, such as on packages, and in verbal form on some attributes, such as in advertising or in a magazine like Consumer Reports. In such situations, consumers may, in effect, weigh information in a particular form more heavily than information in some other form. Relative usage of one form of information versus another may best be captured using a within subjects rather than a between subjects approach. In a within subjects approach where consumers are presented with information in different forms such as both numerical and verbal forms, as opposed to a between subject approach, these two forms of information would be in direct competition. Therefore, these two forms of information could be directly compared in terms of the degree to which consumers use one form of information more than the other. Moreover, a within subjects approach is also realistic in representing many situations where information is available to consumers in both forms from a variety of sources.

The basic proposition tested here is that, when information on some attributes of a brand is presented verbally and information on other attributes is presented numerically, consumers may tend to use the verbal nutrition information to a greater degree than the numerical nutrition information. This proposition was evaluated by generating and testing several hypotheses.

Hypotheses

Several hypotheses were generated and tested to investigate whether verbal nutrition information is used to a greater degree than numerical nutrition information. Using a setting where subjects are exposed to nutrition information on several attributes for several brands and then complete several tasks, hypotheses about the weight given to nutrition information in judgments of healthiness of brands, accuracy of recall of nutrition information, and accuracy of brand ratings along attributes were developed and tested.

The first hypothesis was based on the rationale that, if verbal information is used to a greater extent than numerical information, greater weight would be given to such information in judgments of healthiness of brands.

H1: Nutrition information will be given greater weight in judgments of healthiness when it is presented verbally rather than numerically.

The next hypothesis was based on the rationale that, if verbal nutrition information is used to a greater degree than numerical nutrition information, such usage would be reflected in more accurate subsequent recall of verbal when compared to numerical information. Therefore, greater usage of verbal nutrition information was expected to lead to more accurate subsequent recall of such information.

H2: Recall of nutrition information will be more accurate when it is presented verbally rather than numerically.

Similarly, greater usage of verbal information was also expected to lead to more accurate subsequent ratings of brands along attributes, the basis for H3.

H3: Ratings of attributes based on nutrition information will be more accurate when it is presented verbally rather than numerically.

These hypotheses were tested in an experiment described below.

EXPERIMENT

Overview of Design

The procedures were similar to those used by Viswanathan (1994) in terms of the overall design and stimulus materials with a key exception. The experiment manipulated numerical versus verbal information within subjects in order to directly compare usage of these two types of information by individual respondents. Therefore, nutrition information was presented verbally on some attributes and numerically on other attributes. Subjects were exposed to nutrition information on several fictitious brands along several attributes for a product category with instructions to rate the healthiness of each brand. This was followed by judgments of healthiness of each brand (to test H1), and then ratings of brands along attributes followed by recall of brand information (to test H2 & H3). Furthermore, three groups of subjects were used; a group with no summary information for numerical information (the 'no-summary' condition), a group where the median of values or magnitudes of all available brands on an attribute was provided with numerical information (the 'average' condition), and a group where the maximum and minimum values of all available brands on an attribute was provided with numerical information (the 'range' condition). Viswanathan (1994) used a similar design and showed that summary information in the form of a range or an average facilitate the usage of numerical nutrition information. Such a design was used to compare verbal information to numerical information provided with summary information that has been shown to facilitate its interpretation.

Stimulus Materials

The product category, breakfast cereals, with four attributes (calorie content, sodium content, fiber content, and sugar content) was chosen from Consumer Reports (1990). This product category has several attributes that have implications for healthiness judgments, and has been used in past research (cf., Levy et al. 1991). The brand information presented to subjects is shown in Figure 1. Information on cereals was presented numerically for two attributes (i.e., calorie content and sodium content) and verbally for the other two attributes (i.e., fiber content and sugar content). Therefore, the mode of presented information was manipulated within subjects in order to provide comparisons between numerical and verbal information. Four fictitious brands were used. The highest value, lowest value, 75th percentile value, and 25th percentile value of all brands listed in Consumer Reports (1990) were chosen and assigned to each brand for each attribute presented numerically. This was in order to cover the range of possible values on each attribute and employ an equal number of brands that were above or below the median value of all brands in the market place. The labels 'very low', 'low', 'high', and 'very high' were used for the attributes presented verbally.

H1 predicts that greater weight would be given to verbal information when compared to numerical information. To test H1, the relative healthiness of brand information presented in numerical (versus verbal) form was manipulated between brands to assess the weight given to numerical (versus verbal) information in making judgments of healthiness of brands. Relative healthiness was manipulated by providing information on each attribute that was either above or below the median value for brands along that attribute based on Consumer Reports (1990). Moreover, above and below median values for healthiness were decided on the basis that lower fat content, lower sugar content, lower sodium content, and higher fiber content were desirable for healthiness similar to past research (cf., Levy et al. 1991). These relationships were also suggested to subjects in the instructions. The assignment of specific magnitudes or values to brands of breakfast cereals were such that, on two attributes presented numerically (calorie content and sodium content), two brands were below the median on healthiness (see Figure 1 where Brands C & D, referred to as 'verbally healthy' (i.e., 'numerically unhealthy'), had above median calorie content and above median sodium content) and two brands were above the median on healthiness (see Brands A & B in Figure 1, referred to as 'numerically healthy' (i.e., 'verbally unhealthy')). However, on the two attributes that were presented verbally (fiber content and sugar content), the assignment was reversed so that two brands that were above the median on healthiness on numerical attributes were below the median on healthiness on verbal attributes (i.e., below median fiber content and above median sugar content) and vice versa. Therefore, ratings of healthiness of verbally healthy (i.e., numerically unhealthy) versus numerically healthy (i.e., verbally unhealthy) brands were used as indicators of the weight given to numerical versus verbal information. Following the presentation of nutrition information for each brand, judgments of healthiness of brands were collected to test H1. H2 and H3 relate to the accuracy of recall and the accuracy of brand ratings along attributes, respectively. A rating task was used where respondents were asked to rate each of the four brands along each attribute. Subjects also completed a recall task where they were asked to write down brand information that they remembered.

FIGURE 1

NUTRITION INFORMATION PRESENTED IN EXPERIMENT

Procedures

90 students at a midwestern university participated in the experiment with 30 students being assigned to each of the conditions based on the type of summary information. The experiment was administered using a questionnaire. Subjects were familiarized with the product category of breakfast cereals, and informed about the attributes on which information would be presented and how information would be conveyed along those attributes. Subjects were also informed that the information presented was based on Consumer Reports and had a high degree of accuracy, in order to minimize discounting of information due to factors such as credibility. They were also instructed that "high fiber content, low sugar content, low sodium content and low calorie content are generally considered as being good for health" and familiarized with the fictitious brand names. Additional instructions describing these two types of summary information were provided for the groups in the 'average' and 'range' conditions, using gas mileage of automobiles as an example.

Subjects were exposed to information on a brand of breakfast cereal on the four attributes mentioned above and then asked to rate the brand on several scales which were presented at the bottom of the same page of the questionnaire. Subjects completed four 5 point scales for each brand relating to the healthiness (5 point scale end-anchored not at all healthy - very healthy), nutrition content (5 point scale end-anchored not at all nutritious- very nutritious), liking (5 point scale end-anchored not at all - very much), and likelihood of purchase (5 point scale end-anchored very low - very high) of the brand. This was followed by a similar procedure for the other three brands, each presented on a different page. At the bottom of each page, subjects were instructed not to turn to a previous page in order to prevent direct comparisons across brands. Next, subjects completed five point rating scales labeled Very low - Very high where they rated each brand on calorie content, fiber content, sugar content, and sodium content. Next, subjects performed a free recall task where they were instructed to write down the information they remembered (i.e., brand name, attribute name, and value), and to write the value in any form in which it came to mind (i.e., in numerical or in verbal form). Finally, importance ratings for each attribute were collected using 7 point scales labeled Not at all important - Very important.

Results

Results of Ratings of Healthiness - H1. Mean healthiness ratings were computed for each subject for the two numerically healthy (i.e., verbally unhealthy) brands and also for the two verbally healthy (i.e, numerically unhealthy) brands. Responses to the scale on healthiness of brands (5 point scale end-anchored not at all healthy - very healthy) were used for this analysis. A 3 (type of summary information; no-summary, average, and range; between subjects) by 2 (numerically healthy (i.e., verbally unhealthy) versus verbally healthy (i.e., numerically unhealthy) brands; within subjects) ANOVA was performed on these mean healthiness ratings. A significant main effect of numerically healthy versus verbally healthy brands was obtained (F (1,87)=54.26; p<.001). Verbally healthy brands had higher healthiness ratings than numerically healthy brands (3.73 versus 2.74 on a 5 point scale). Stated differently, verbally unhealthy brands had lower healthiness ratings than numerically unhealthy brands. The pattern of results provide support for H1. A non-significant interaction was obtained between type of summary information and numerically healthy versus verbally healthy brands suggesting that the advantage for verbal information persists even when numerical information is presented with summary information to facilitate its interpretation.

Results of Recall - H2. Subjects in the recall task were instructed to recall information in any form they preferred leading to numerical and verbal recall of information that was numerical at presentation, and verbal and numerical recall of information that was verbal at presentation. The proportion of accurately recalled items for each of these forms of recall was computed for each subject. Accurate recall was computed separately using two different criteria; a lenient criterion and a strict criterion. Using the lenient criterion, recall was considered accurate when a recalled item was within one scale-point on either side of the original item based on a five point scale of the 0th, 25th, 50th, 75th, and 100th percentile value on an attribute (e.g., recall of "low" sugar content for a brand as "very low" or "neither low nor high" was considered as being accurate; recall of "125" calories for a brand (i.e., the highest value) as "very high" or "high" was considered as being accurate). Such a criterion for accuracy was used to allow for individual differences in the manner in which subjects translate numerical labels and also to allow for approximate rather than exact recall. Accurate recall using the strict criterion required a recalled item to be identical to the original item based on a five point scale of the 0th, 25th, 50th, 75th, and 100th percentile value on an attribute (e.g., recall of calorie content for a brand of "125" calories (i.e., the highest value) as "very high", i.e., the verbal equivalent on the five point scale described above, was considered as being accurate). The recall data was examined to identify accurately recalled items separately using each criterion and scores were assigned to each subject based on the proportion of all items that were accurately recalled.

A 3 (type of summary information; no-summary, average, and range) by 2 (mode at exposure; numerical versus verbal) by 2 (mode at recall; numerical versus verbal) factorial ANOVA was run on the proportion of accurate recall using the lenient criterion. A significant main effect was obtained for the mode of information at exposure (F (1, 76)=14.02; p<.001) with higher accuracy for verbal information (.69 versus .58 for verbal versus numerical information). The pattern of results provides support for H2. The interaction between mode at exposure and type of summary information was non-significant suggesting that the advantage for verbal information occurred even when numerical information was presented with summary information to facilitate its interpretation.

A 3 (type of summary information; no-summary, average, and range) by 2 (mode at exposure; numerical versus verbal) by 2 (mode at recall; numerical versus verbal) factorial ANOVA was run on the proportion of accurate recall using the strict criterion. A significant main effect was obtained for the mode of information at exposure (F (1, 76)=17.51; p<.001) with higher accuracy for verbal information (.36 versus .26 for verbal versus numerical information). Again, the pattern of results provides support for H2. The interaction between mode at exposure and type of summary information was non-significant as with analyses using the lenient criterion.

Results of Ratings along Attributes - H3. Accuracy of brand ratings along attributes used to test H3 were also computed separately using two different criteria; a lenient criterion and a strict criterion. Accurate rating using the lenient criterion required a rating on the five point scale to be within one scale-point on either side of the original item based on a five point scale of the 0th, 25th, 50th, 75th, and 100th percentile value on an attribute. Accurate rating using the strict criterion required a rating to be identical to the original item based on a five point scale of the 0th, 25th, 50th, 75th, and 100th percentile value on an attribute. The data was examined to identify accurately rated items separately using each criterion and scores were assigned to each subject based on the proportion of all items that were accurately rated.

A 3 (type of summary information; no-summary, average, and range) by 2 (mode at exposure; numerical versus verbal) factorial ANOVA was run on the proportion of accurate ratings using the lenient criterion. A significant main effect was obtained for the mode of information at exposure (F (1, 87)=7.52; p<.01) with a higher accuracy for verbal information (.85 versus .79 for verbal versus numerical information). The pattern of results provides support for H3. The interaction between mode at exposure and type of summary information was non-significant suggesting that the advantage for verbal information occurred even when numerical information was presented with summary information to facilitate its interpretation.

A 3 (type of summary information; no-summary, average, and range) by 2 (mode at exposure; numerical versus verbal) factorial ANOVA was run on the proportion of accurate ratings using the strict criterion. A significant main effect was obtained for the mode of information at exposure (F (1, 87)=7.64; p<.01) with a higher accuracy for verbal information (.40 versus .34 for verbal versus numerical information). The pattern of results provides support for H3. The interaction between mode at exposure and type of summary information was non-significant as with the analyses based on the lenient criterion.

Discussion of Results

All the hypotheses were supported by the findings. The pattern of results suggest that the provision of nutrition information in a verbal form when compared to a numerical form leads to several advantages in terms of weight given to brand information in making judgments of healthiness, accuracy of subsequent ratings of brands based on attribute information, and accuracy of recall of brand information. These advantages appear to persist even when numerical information is presented with summary information to facilitate its interpretation.

An alternate explanation of the findings in light of the design is that the two attributes presented verbally were more important than the two attributes presented numerically, hence the greater usage of verbal nutrition information. However, in choosing attributes for the study, attributes that were of relevance in making judgments of healthiness were chosen, therefore, all four attributes were likely to be of comparable importance. Importance ratings for each attribute were collected at the end of the study using 7 point scales labeled Not at all important - Very important. These ratings were examined to explore the alternate explanation and appeared to be comparable for the four attributes; 4.02 for calorie content (presented numerically), 5.08 for sugar content (presented numerically), 5.16 for sodium content (presented verbally), and 5.34 for fiber content (presented verbally). A possible exception here is calorie content (presented numerically) which has a somewhat lower rating. However, it is unlikely that this difference in attribute importance in itself influenced the results in terms of a consistent advantage for verbal information across several variables.

GENERAL DISCUSSION

The primary objective of this paper was to compare verbal versus numerical presentations in facilitating the usage of nutrition information. Verbal nutrition information, due to its descriptive nature, was argued to be used to a greater degree than numerical nutrition information, which requires a reference point in order to be interpreted. Hypotheses were developed and tested to evaluate this proposition. The results of an experiment suggest that verbal information may have several advantages in terms of weight given to brand information in making judgments of healthiness, accuracy of subsequent ratings of brands based on attribute information, and accuracy of recall of brand information. These advantages appear to persist even when numerical information is presented with summary information to facilitate its interpretation. This research demonstrates several advantages for verbal when compared to numerical nutrition information.

Limitations of this study include the artificial nature of the experiment. In more realistic settings, nutrition information may be available to consumers on a larger number of attributes than were used in the experiment. Furthermore, additional variables such as comparisons between brands on nutrients need to be studied. The assignment of certain attributes to numerical versus verbal conditions is a potential weakness of this design, although the importance of the attributes appeared to be comparable. The composition of the sample also restricts the generalizability of the findings.

In interpreting the findings of this research, the conditions under which an advantage for verbal information were obtained need to be examined. In the experiment described here, all information on an attribute was either verbal or numerical. Furthermore, judgments of healthiness of brands were made while subjects were exposed information. Although not limitations in a strict sense, these conditions should be noted in interpreting the results of this study. An understanding of the conditions under which there is an advantage for one form of information versus another could provide insight into the processing of these two forms of information. Different results may be obtained if, say, all information on an attribute is not of the same form, or if the task at exposure to information is different.

The limitations described above notwithstanding, this study has important implications for consumer research, both in the area of nutrition information and more generally in the area of product information. This research supplements past consumer research (Scammon 1977; Viswanathan 1994) that has shown advantages for verbal when compared to numerical information. This study also directly compares the use of numerical versus verbal information by manipulating the form of information within subjects. Verbal nutrition information, due to its descriptive nature, appears to be used to a greater degree than numerical nutrition information. Therefore, a key characteristic of product information that appears to facilitate its usage is the degree to which it directly conveys or describes the location of a brand along an attribute. This dimension of descriptiveness could be used to understand the processing of different forms of information by consumers. For example, numerical ratings on a generic scale, such as those used in Consumer Reports, are more descriptive than numerical information on a unit of measurement, such as calories. Such ratings convey the location of a brand more directly because they can be easily interpreted using the end points of the scale. On the other hand, % of Daily Value is similar to numerical information on a unit of measurement because the relative location of a brand on an attribute is not directly conveyed by this form of information.

For practitioners, this study suggests several advantages in the use of verbal information. From the perspective of public policy makers, the use of verbal labels to describe product attributes would require a high degree of preprocessing to develop judgments about highness and lowness on product attributes that could otherwise be made by individual consumers. It may be easier to provide numerical information along with summary information that facilitates its interpretation. Nevertheless, the use of verbal information to supplement numerical information may facilitate the processing of nutrition information by consumers (cf., Levy et al. 1991). Furthermore, the evidence suggesting that verbal information may be used to a larger degree than numerical information highlights the importance of norms for the use of specific verbal labels such as "light" and "green" by manufacturers. For marketers, this research suggests that consumers may tend to give greater weight to verbal when compared to numerical information under certain conditions.

Several lines of future research are suggested by this study. One line of future research should focus on the processes involved in using numerical versus verbal information. The conditions that lead to greater use of one form of information versus the other need to be studied. Such research may provide insight into the processing of these two forms of information. Further research is needed to understand how verbal versus numerical information is used in decision making, and how learning and memory is influenced by these two forms of information. Field studies using verbal versus numerical nutrition information would also provide insight into the effects of variables that exist in realistic settings. In conclusion, the study of verbal versus numerical presentations of nutrition information provides promising avenues for future consumer research both in the specific area of nutrition information and in the more general area of product information.

REFERENCES

Consumer Reports (1990), v. 55, No. 12, 1991 Buying Guide Issue.

Huber, Oswald (1980), "The Influence of Some Task Variables on Cognitive Operations in an Information-Processing Decision Model," Acta Psychologica, 45, 187-196.

Levy, Alan S., Odonna Mathews, Marilyn Stephenson, Janet E. Tenney, and Raymond E. Schucker (1985), "The Impact of a Nutrition Information Program on Food Purchases," Journal of Public Policy and Marketing, 4, 1-13.

Levy, Alan S., Sara B. Fein, and Raymond E. Schucker (1991), "Nutrition Labeling Formats: Performance and Preference," Food Technology, July, 116-121.

Moorman, Christie (1990), "The Effects of Stimulus and Consumer Characteristics on the Utilization of Nutrition Information," Journal of Consumer Research, 17 (3), 362-374.

Muller, Thomas E. (1985), "Structural Information Factors which Stimulate the Use of Nutrition Information: A Field Experiment," Journal of Marketing Research, 22, 143-157.

Russo, J. Edward, Richard Staelin, Catherine A. Nolan, Gary J. Russell, and Barbara L. Metcalf (1986), "Nutrition Information in the Supermarket," Journal of Consumer Research, 13, 48-70.

Scammon, Debra L. (1977), "'Information Load' and Consumers," Journal of Consumer Research, 4 (December), 148-155.

Venkatesan, M., Wade Lancaster and Kenneth W. Kendall (1986), "An Empirical Study of Alternate Formats for Nutritional Information Disclosure in Advertising," Journal of Public Policy and Marketing, 5, 29-43.

Viswanathan, Madhubalan, "The Influence of Summary Information on the Usage of Nutrition Information," Journal of Public Policy and Marketing, 13 (1), 48-60.

Viswanathan, Madhubalan, and Terry Childers (1992). The Encoding and Utilization of Magnitudes along Product Attributes: An Investigation Using Numerical and Verbal Information, Unpublished Manuscript.

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