Health claims and consumers’ behavioral intentions: The case of soy-based food

Food Policy 36 (2011) 480–489

Contents lists available at ScienceDirect

Food Policy journal homepage: www.elsevier.com/locate/foodpol

Health claims and consumers’ behavioral intentions: The case of soy-based food q Wanki Moon a,⇑, Siva K. Balasubramanian b, Arbindra Rimal c a

Department of Agribusiness Economics, Southern Illinois University, Carbondale, IL 62901, USA Stuart School of Business, Illinois Institute of Technology, Chicago, IL 60616, USA c Department of Agriculture, Missouri State University, Springfield, MO, USA b

a r t i c l e

i n f o

Article history: Received 14 September 2009 Received in revised form 27 February 2011 Accepted 10 May 2011 Available online 8 June 2011 Keywords: FDA approval Health claim Soy foods Health benefits Behavioral intentions

a b s t r a c t This research evaluates the impact of two soy-specific health claims (highlighting FDA approval along with scientific results and simply describing scientific results) on stated behavioral intentions toward soy-based food using a survey administered by Ipsos-Observer to a nationally representative web panel in the summer of 2007. Our research design randomly assigned respondents to a health claim. Three ordered probit models (non-soy users; infrequent soy users; regular soy users) show that non-soy users and infrequent soy users who were exposed to either FDA health claim or general health claim are significantly more likely to eat soy-based food products. FDA or general health claim, however, did not change the behavioral intentions of regular soy users. These results suggest that soy consumption status moderates the impacts of health claims on behavioral intentions. However, the impact of FDA health claim did not differ from that of general health claim, indicating that the word ‘FDA’ did not add any additional information to consumers beyond the general health claim. Ó 2011 Elsevier Ltd. All rights reserved.

Introduction Research has established that consumers’ knowledge of the linkage between diet and health is a significant determinant of their dietary choices (Chern et al., 1995; Variyam et al., 1998; Chern, 2002; Brown and Schrader, 1990). Recent studies confirmed this finding specifically in the context of soy-based foods. For example, Moon et al. (2005) showed that consumers’ perceived health benefits of soy foods significantly increase the likelihood of consuming soy food as well as how often soy is consumed. Wansink et al. (2005) suggest that soy consumption is more likely when individuals are aware of soy attributes that are personally relevant: i.e., individuals are more likely to consume soy foods when they possess attributerelated knowledge and consequence-related knowledge. The soy-related studies above were motivated in large part by the 1999 decision of the Food and Drug Administration (FDA) allowing food manufacturers to use health claims on foods with at least 6.25 g of soy protein per serving. From a scientific perspective, this decision was based on medical and clinical data showing that 25 g of soy protein a day significantly lowered both totalcholesterol (TC) and low-density-lipoprotein cholesterol (LDC) in the bloodstream – the two most important modifiable risk factors for coronary heart disease (Anderson et al., 1995; Hasler, 2002;

q The authors would like to acknowledge financial support for this research from Illinois Missouri Biotechnology Alliances (IMBA). ⇑ Corresponding author. Tel.: +1 618 453 6741; fax: +1 618 453 1708. E-mail address: [email protected] (W. Moon).

0306-9192/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.foodpol.2011.05.001

Devine, 2002). Cardiovascular heart disease is the number one cause of deaths in the US. Other clinical research has shown that soy foods provide health benefits relative to osteoporosis and cancer (Messina and Barnes, 1995), lower blood pressure and lowered blood levels of triglycerides. As a result, soy foods have acquired a growing prominence in the American diet. Retail soyfood sales in the US amounted to $4 billion in 2005 (Soyatech Inc. and Spins Inc., 2005) and are expected to increase to $8.7 billion in 2009 (Freedonia Group, 2005). Strategic alliances among food companies to introduce soy products into the market place, updated dietary guidelines by the USDA and Department of Health and Human Services that listed soyfoods as means to meet the dietary recommendations, and USDA approval of soy as a meat substitute in school lunch program are among other factors contributing to the growth of soyfood market (Pszczola, 2000). Building on the established linkage between soy health knowledge and soy consumption behavior, questions of research interest from a behavioral perspective include: (i) do soy health claims effectively increase consumers’ knowledge of health benefits and/or alter consumer attitudes and behavior, (ii) which information format (length, wording) is most effective if soy health claims impact consumer behavior, (iii) if a soy health claim showcases FDA’s approval of the health benefits of soy protein, does that increase the claim’s effectiveness in terms of consumers’ attitudes and behavior, and (iv) does the degree of consumers’ involvement in soy food moderate the impact of health claims on their behaviors and attitudes. Recent research affords some insight into the first two questions above. Wansink et al. (2000) show that, although the ‘‘soy’’ label on a package negatively influenced consumers, when combined with a

W. Moon et al. / Food Policy 36 (2011) 480–489

health claim, it improved attitudes among consumers who are health-conscious, natural food lovers, or dieters. Additionally, Wansink (2003) found that combining short health claims on the front of a package with full health claims on the back of the package increased the believability of the health claim. The third question centers around the source credibility effect, if any, that is associated with the FDA. Wansink and Cheney (2005) emphasize two general conditions for leveraging FDA approved health claims (increasing consumers’ awareness that the product carrying the claim has the target nutrient, and ensuring that the target nutrient provides a health benefit that is relevant to consumers), but research on the ‘value added’ by FDA to soy claims that showcase FDA approval awaits exploration. This study attempts to provide new insights into these questions. Using a survey administered to a nationally representative webpanel, we evaluate the impact of health claims on consumers’ behavioral intentions toward soy-based food after controlling for individual characteristics including psychological and socio-demographic profiles. Two factors broadly motivate this study’s focus on behavioral intentions with respect to soy. First, given the low penetration of soy foods into the American diet (Wansink and Chan, 2000) and the documented scientific evidence on the health benefits of soy, efforts to increase soy consumption are likely to enhance consumer welfare. Of particular interest here are consumers who are not participating in the soy market or consume soy food only sporadically. Second, it is very important to enhance our understanding of how consumers are influenced by health claims in soy products. As the next section indicates, a rich body of research literature focuses on the consumer welfare implications of health claims in food products. However, the impact of health claims in soy products remains to be explored from a research perspective.

Nutrient/health claim information, and consumer behavior Growing demand for health attributes of foods coupled with accumulating scientific knowledge on the link between diet and disease have stimulated a broad range of economic and behavioral research on the role of health claims and food labels in consumers’ food choices (e.g., Goland and Unnevehr, 2008; Caswell and Padberg, 1992; Nayga et al., 1998; Unnevehr and Hasler, 2000; Mojduszka and Caswell, 2000; Roosen et al., 2009). First, addressing health concerns has become a central theme of marketing efforts within the food industry (e.g., new product development, market segmentation, advertising or promotion). Research indicates that the use of health claims (associating certain nutrients to the reduced risks of specific diseases) and other content-related claims on food products have increased over time (Caswell et al., 2003). Second, there is evidence (Roe et al., 1999) that the presence of a health or nutrition-content related claim may induce individuals to either stop their information search or to place greater reliance on the claim than on information in the ‘Nutrition Facts’ panel. In some instances, claims even generate a halo effect (rate a food higher on attributes not mentioned in the claim) or a magic bullet effect (ascribe inappropriate health benefits to a product). Third, Burton and Creyer (2004) note that a claim may have a positive influence on disease risk perceptions even when it is not specific to a disease. Fourth, Garretson and Burton (2000) suggest that consumers may not be consistent in processing claim information across nutrients. For example, they appear to be less sensitive to incongruencies between claim and ‘Nutrition Facts’ information for fiber, as opposed to fat. More recently, France and Bone (2005) asserted that product labels may not directly engender product-specific beliefs about claim-articulated benefits. Rather, consumers may use various biasing filters (e.g., trust in government, health motivation) to process claims that appear on

481

a product’s label. Overall, using claims to underscore health or nutrition attributes of food products raises questions about their truthfulness and/or unexpected consequences. From a consumer welfare perspective, it is clear that the regulation of health claims on food products has emerged as a major public policy concern. Two related questions are addressed next: How is such regulation structured? At a fundamental level, do consumers benefit from health claims?

Regulation of health claims While some form of regulation is usually necessary to ensure the truthfulness of the information disseminated, this oversight responsibility historically rested with the FDA (Food and Drugs Administration – see Calfee and Pappalardo, 1991; France and Bone (2005) for excellent reviews about this). Initially however, the concern over potentially deceptive messages prompted the FDA to ban all health claims on food products. Nevertheless, this extreme regulatory step was seen by the food industry and public interest groups as impeding consumers’ interests. Later (early 1980s), the accumulation of scientific evidence linking dietary fiber to colon cancer emboldened the breakfast cereal industry to use health claims on their products. For example, Kellogg promoted All-Bran cereal with health claims stating that the high-fiber product could reduce the risk of colon cancer (Ghani and Childs, 1999). Clearly, this approach violated then-prevailing FDA rules banning health claims on food products. Following a policy review, the ban against health claims was suspended in 1985. Subsequently, the NLEA streamlined regulations [21 CFR 101.14] in a manner that reinforced FDA’s authority by identifying specific health claims (and related circumstances) that are allowed by FDA. In other words, FDA had both the authority and responsibility to institute rigorous scientific review before establishing a health claim that linked any nutrient to a disease. More recently however, sections 303 and 304 of the FDA Modernization Act of 1997 (FDAMA) permit health claims that are based on ‘‘current, published, authoritative statements from certain federal scientific bodies, as well as from the National Academy of Sciences’’ (see FDA, 1998). These FDAMA sections formally allow petitions that propose new health claims. The aim here is to expedite the process by which the scientific basis for any new health claim gets established. The preceding discussion lends context to FDA’s approval of the health claim for soy foods in October 1999, in response to a petition submitted by Protein Technologies International. Given the evolutionary changes in FDA policy toward health claims, it is clear that the focus of health claim regulation rests more on establishing the scientific basis of the claim and less on any perceived FDA endorsement value. However, it is useful to empirically determine if a health claim that showcases its official approval by FDA in the context of scientific findings (e.g., the FDA health claim on the left side of Table 1) generates a higher impact on consumers than a health claim that merely states the scientific findings (e.g., the general health claim on the right side of Table 1). Following the source credibility literature (Jain and Posavac, 2001; Sternthal et al., 1978; Wiener and Mowen, 1986) that suggests that a health claim will be deemed as more acceptable if its source is perceived as more credible, our research goal here is to compare consumer outcomes following exposure to the two health claim formats in Table 1 to assess if FDA’s approval adds greater credibility to the health claim. Previous research has focused on manufacturer credibility with respect to claims on food products (Keller et al., 1997), but the potential role of FDA approval remains to be explored from a credibility angle. Hence, we present a research question next instead of a hypothesis:

482

W. Moon et al. / Food Policy 36 (2011) 480–489

Table 1 FDA and general health claims. (1) FDA approved health claim

(2) Health claim explaining scientific research

The Food and Drug Administration (FDA) officially confirmed the health benefits of soy-based foods with a 1999 ruling that food manufacturers can claim that 25 g of soy protein a day as a part of diet low in saturated fat and cholesterol may reduce the risks of coronary heart diseases. This ruling is based on the scientific finding that this soy protein dosage reduced blood cholesterol levels by 9.3% and the risk of coronary heart disease by 18.6%

Scientific research shows that 25 g of soy protein a day as a part of a diet low in saturated fat and cholesterol may reduce the risk of heart disease. This soy protein dosage reduced blood cholesterol levels by 9.3% and the risk of coronary heart disease by 18.6%

RQ1: Does FDA approval add any further information or credibility to a health claim? Consumers’ benefits from health claims Ippolito and Mathios (1991) demonstrated that using health claims significantly increased consumer knowledge of the fibercancer relationship, increased consumption of fiber cereal, and stimulated product innovations. Focusing on the shift in FDA policy toward health claim regulation, Ippolito and Mathios (1994, 1995) showed that the consumption of fats and saturated fats decreased faster after 1985 when compared to the 1977–1985 period. Mathios (1998) investigated the market for cooking oils before and after the implementation of the NLEA that prohibited health claims for foods whose content per serving for a specified nutrient exceeded a prescribed limit (Ford et al., 1996). Since cooking oils contain more fats than the prescribed threshold, the NLEA eliminated health claims for cooking oils. Previously, cooking oils lower in saturated fats and higher in monounsaturated fats were shown to be superior to other oils from a heart-health perspective. Using supermarket scanner data, Mathios (1998) showed that the implementation of the NLEA caused consumers to shift purchases toward cooking oils higher in saturated fat and lower in monounsaturated fat. Collectively, the studies reviewed earlier strongly attest to the value of health claims on food products as an information resource for consumers. Therefore, consumers who were exposed to health claims extolling the beneficial effects of soy-based foods are likely to be more willing to consume soy foods. We propose that: H1. Consumers exposed to either the FDA health claim or general health claim are more likely to exhibit positive behavioral intentions toward soy foods when compared to consumers not exposed to health claims. Behavioral intentions expressed in response to the three stimuli may differ across respondents depending on their status on soybased food consumption. For example, non-soy users may be less likely to exhibit positive behavioral intentions when compared to regular or infrequent users because they are likely to have a relatively lower interest or motivation in soy-based foods. This point is in line with the role of consumer involvement in explaining behavioral outcomes that has been extensively researched in the literature (e.g., Zaichkowsky, 1985; Gordon et al., 1998; Beharrell and Denison, 1995; Verbeke and Vackier, 2004) Accordingly, we propose that: H2. Behavioral intentions depend upon the degree of consumers’ engagement in the soy market. To evaluate this conjecture, we divide the sample into three categories: (i) regular users of soy foods, (ii) infrequent users, and (iii) non-users. We used a screening question to assign survey participants to appropriate questions of the following three types of behavioral intentions: (i) would you be willing to increase your soy consumption? (directed toward regular soy users) (ii) Would you be willing to include soy foods regularly

in your diet? (directed toward infrequent users) (iii) Would you be willing to try soy foods? (directed toward non-soy users). The idea of dividing behavioral intentions into these three types is in line with the findings of Wansink et al. (2000) and Moon et al. (2005) that illustrate the importance of categorizing consumers based on their soy consumption status when analyzing soy consumption behavior. Lastly, the above soy consumption status may moderate the impacts of health claims on behavioral intentions. More specifically, we can hypothesize that the impact of FDA treatment on behavioral intentions would be larger in the case of non-soy users or infrequent users compared to regular users, given that the FDA health claims are less likely to be new information to regular soy users and that they may have little room in their diets to increase consumption of soy foods. H3. Soy consumption status moderates the impact of health claims on behavioral intentions.

Theoretical and empirical models Methodologically, this study relies on Ajzen and Fishbein’s (1977, 2005) theory of planned behavior to assess the effect of health claims on consumers’ behavioral intentions toward the consumption of soy-based food products. Specifically, we model behavioral intentions toward soy-based food products as functions of the following four sets of variables along with health claims: (i) perceived attributes about soy food including health benefits, taste, convenience, and price; (ii) health knowledge and nutritional awareness, (iii) socio-economic and demographic profiles, and (iv) soy consumption status as reflected in three groups (non-soy users; infrequent users; and regular users). Figs. 1 and 2 depict the model framework to be estimated later. In particular, the model indicates that: (a) behavioral intentions may differ by soy consumption status and (b) such consumption status may moderate the impact of health claims on behavioral intentions. We conjecture that regular soy users are more likely to be aware of the health benefits of soy foods, so they may be less responsive to health claims when compared to non-soy users. Under the circumstances, a FDA approved health claim may not offer new information to regular soy users. Specification of regression models We draw upon Figs. 1 and 2 to link behavioral intentions to a set of explanatory variables encompassing perceived attributes of soy foods, health related factors, socio-demographic profiles, and the two treatments of health claims. The equations below represent regression models.

Yi ¼ b1 X1 þ e1i i ¼ non-soy users

ð1aÞ

Yj ¼ b2 X2 þ e2i j ¼ infrequent soy users

ð1bÞ

Yk ¼ b3 3 þ e3i k ¼ regular soy users

ð1cÞ

W. Moon et al. / Food Policy 36 (2011) 480–489

Health Knowledge Nutritional Awareness

Perceptions about Attributes Of Soy Foods Health Benefits Price Taste Convenience

Socio-demographics: Income Education Age Gender Ethnic Background

483

Soy Consumption Status Non-soy users Infrequent users Regular users

Behavioral Intentions Toward Soy Foods

Stimuli FDA Health Claim General Health Claim

Fig. 1. Theoretical model explaining consumers’ behavioral intentions toward soy food.

Predictor: FDA and General Health Claims

Moderator: Soy Consumption Status

Outcome Variable: Behavioral Intentions

Predictor * Moderator

Fig. 2. Moderating roles of soy consumption status in the effects of health claims on behavioral intentions (adapted from Baron and Kenny, 1986).

where Yi, Yj, and Yk represent behavioral intentions of non-soy users, infrequent soy users, and regular soy users, respectively. Specifically, behavioral intention of non-soy users is defined as willingness to try soy-based food products; for infrequent soy users it is defined as willingness to include soy foods regularly in their diet; and for regular soy-users it is defined as willingness to increase the consumption of soy foods. X1, X2, and X3 are vectors of variables explaining the behavioral intentions of non-soy users, infrequent soy users, and regular soy users, respectively. We assume that X = [X1 = X2 = X3], implying that the same set of variables impact the behavioral intentions of the three groups of consumers. However, estimating regression model for each of the three groups of consumers will permit the variables in X to have differential effects across the three groups of consumers. Let X be partitioned into four matrices X = [PA, DG, HK, HC]. The matrix (PA) represents respondents’ perceptions about the attributes of soy-based foods (health-promoting attribute, taste, convenience, price) which is hypothesized to influence the behavioral intentions of respondents. For example, if respondents are predisposed to have a favorable perception about the taste of soy-based food, they would be likely to exhibit more positive behavioral intentions toward soy food. Similarly, respondents who perceive soy food as possessing

health-promoting attribute are more likely to show positive intentions toward consuming soy food. The matrix (HK) shows healthrelated factors including health knowledge and nutritional awareness. The matrix (DG) represents socio-demographic profiles including income, ethnic background, age, and education which can have relevance in explaining respondents’ behavioral intentions toward soy-based food. The matrix (HC) includes FDA and general health claims. These variables will enter as dummies and ‘no statement’ case is dropped and used as a reference group. In sum, three separate equations are estimated for non-soy users, infrequent users, and regular users. Subsequently, we estimate an aggregate model combining all three groups as described below.

Yi ¼ bX þ cS þ hM þ ei i ¼ all respondents

ð2Þ

This aggregate model includes two additional vectors (S and M) compared to Eqs. (1a), (1b), and (1c): (i) S = two binary dummies for soy consumption status (regular and infrequent users), and (ii) M = four interactions terms (FDA health claim⁄infrequent users = Interaction#1; FDA health claim  Regular users = Interaction#2; General health claim  Infrequent users = Interaction#3; General health claim  Regular users = Interaction#4). The

484

W. Moon et al. / Food Policy 36 (2011) 480–489

aggregate model poses a problem because the three dependent variables that were worded differently across the three groups of consumers with respect to the status of soy consumption were merged into a single dependent variable. Yet, the three variables can be viewed as homogeneous in the sense that they measure some facet of the intensity of behavioral intentions in relation to soy-based food products in a continuum from non-soy users to infrequent soy users to regular users. The aggregate model serves two purposes of importance in better understanding the role of soy consumption status in assessing the impact of health claims on behavioral intentions. First, it offers a direct avenue to evaluate the role of soy consumption status in determining behavioral intentions toward soy foods. With the three separate equations for three groups of respondents (non-soy users, infrequent users, and regular users) that were developed earlier, we would not be able to tell directly whether the soy consumption status has any relevance in explaining behavioral intentions. Second, the four interaction terms between the two treatments and soy consumption status could overtly detect potential moderating roles of soy consumption status on the effects of FDA and general health claims on behavioral intentions; i.e., the impacts of the two health claims on behavioral intentions could differ across non-soy users, infrequent users, and regular users. Survey design, sampling, and administration This research is based on a survey instrument consisting of five major sections including: (i) questions measuring perceptions about salient attributes of soy-based food products (PA) including health benefits, taste, convenience, and price, (ii) questions measuring nutritional awareness and health knowledge (HK) with a set of 11 questions, (iii) questions measuring soy food consump-

tion behavior (S), (iv) socio demographic profiles, and (v) experimental design investigating the role of FDA and general health claims (HC) on respondents’ behavioral intentions toward soybased food products (more details on these constructs are provided in the next three sections and Table 2). The survey was administered online by Ipsos-Observer in the summer of 2007. Ipsos-Observer is private consulting firm specializing in consumer research and public opinion polls on socially important issues including tracking trends in food consumption. This firm maintains an on-line panel that consists of 400,000 households. Appropriately stratified by geographic regions, income, education, and age to correspond to the 2000 US Census, a sample of 9000 households were drawn out of the online panel in a manner that is representative of the US population. A total of 3456 panel members returned completed online surveys, resulting in an 38.4% response rate. This type of research based on preexisting panels is known as permission-based survey and permits the researcher to have access to demographic information for non-returners as well as returners. We divided the sample into three categories (regular users of soy foods [n = 731], infrequent soy users [n = 735], and non-users [n = 1987]) with the help of a screening question that guided each participant to appropriately choose one of the following three questions: (i) would you be willing to increase your soy consumption? (ii) Would you be willing to include soy foods regularly in your diet? (iii) Would you be willing to try soy foods? Experimental design assessing the role of health claims The last section represents between-subject design providing the basis for the main analysis in this article. As shown earlier in Table 1, two different versions of the label were created and

Table 2 Description and summary statistics of variables used in estimation. Variable Perceived soy attributes (i) Health benefits Lowering cholesterol Antioxidant Bone mass Menopause (ii) Convenience Convenient Recipes Preparation (iii) Taste (iv) Price Health knowledge Nutritional awareness Fresh fruit Fresh vegetables Less fat Less cholesterol Chronic diseases Salt Calcium Nutrition Socio-demographics Age Education

Description

Mean

Alpha

St. dev.

Soy Soy Soy Soy

4.14 4.74 4.69 4.51

.899

1.51 1.28 1.23 1.63

Soy foods are convenient Recipes are readily available I know how to prepare soy foods I like the taste of soy foods Soy foods are inexpensive Aggregation of correct answers to 11 nutritional questions

3.42 3.71 2.82 3.21 3.45 6.14

.727

1.52 1.62 1.48 1.49 1.73 2.17

I I I I I I I I

4.27 4.19 4.3 4.27 4.41 4.67 4.32 4.81

.875

1.71 1.52 1.57 1.81 1.53 1.64 1.89 1.29

foods foods foods foods

lower cholesterol level act as an antioxidant help retain bone mass are good for women during menopause

eat a lot of fresh fruits eat a lot of fresh vegetables actively try to consume less fat actively try to consume less cholesterol am very concerned about linkages between diet and chronic diseases am concerned about the amount of salt in my diet am concerned about getting enough calcium in my diet am generally concerned about nutrition

41.3 4.8

5.6 12.54

Gender

In years 1 = Grade, 2 = Some high, 3 = High graduated, 4 = Some college, 5 = 2 year college, 6 = 4 year college, 7 = Some post graduate, 8 = Post graduate degree 1 if Male; 0 otherwise

0.48

1.63

Ethnic background White Asian Black

1 if Whites; 0 otherwise 1 if Asian; 0 otherwise 1 if Black; 0 otherwise

0.69 0.15 0.13

Note: Seven-point ratings of agreement (1 = strongly disagree; 7 = strongly agree) were used to measure perceived health benefits, taste, convenience and price, health motivation, and nutritional awareness. Alpha represents Cronbach’s measure of internal consistency.

W. Moon et al. / Food Policy 36 (2011) 480–489

randomly assigned across respondents: treatment (1) shows that the FDA has allowed food manufacturers to use health claims on soy foods and presents scientific research results demonstrating that the intake of 25 g of soy protein can lower the chances of heart diseases by 9.3%; treatment (2) just mentions about the scientific research that shows positive effects of soy protein on heart diseases, hence identical with treatment (1) without the part of FDA approval. In addition, a third version was created to represent treatment where no information was presented. Respondents are asked to state their behavioral intentions given their assignment to one of the three treatments. It is essential to discuss the critical importance of random assignment of health claims in our research design, because the correlation between all the model variables on the right hand side and exposure to the various health claims should be minimal given this random assignment of health claim treatment conditions to respondents. Stated differently, absent random assignment, unobserved differences in preferences for health are likely to influence individual’s exposure to health claims and other measured variables – a consequence likely to bias results unless all these unobserved differences are carefully controlled. It is impossible to control for unobserved differences on several germane variables (or even to measure them) in experimental settings, so random assignment is used to avoid this problem. Measures Dependent variables Dependent variables in our study (willingness to consume soy food regularly; willingness to increase the frequency of soy food consumption; willingness to try soy food) were measured with a five point Likert scale (Definitely would not; Probably would not; Might or might not; Probably would; and Definitely would). Perceived attributes of soy-based foods Four attributes of soy food are considered in our study: price, taste, convenience, and health benefit. The survey instrument incorporated questions measuring respondents’ perceptions of these attributes. Previously validated multi-item scales did not exist for perceived soy attributes. Based on a literature search, we initially identified the following soy attributes that are perceived as important (specific health benefits, convenience, taste and price). Scale items were developed and refined subsequently as recommended by Churchill (1979). Perceptions about soy health benefits were measured as composite index (Cronbach’s alpha = 0.899) based on responses to four statements on a seven-point rating scale (1 = strongly disagree, 7 = strongly agree). Perceived convenience was measured by the sum of three items of questions (Cronbach’s alpha = 0.727). In addition, perceived price and taste were measured with the statements ‘‘Soy foods are inexpensive’’ and ‘‘I like the taste of soy-based foods’’, respectively. General health-related variables Given our study’s focus on the role of health claims in determining behavioral intentions toward soy foods, general health-related factors are expected to be relevant in this research. Specifically, two such factors (health knowledge and nutritional awareness) are considered. Moorman and Matulich (1993) define health knowledge as the extent to which consumers have enduring health-related cognitive structures. The health knowledge construct was measured by adapting an instrument drawn from Moorman and Matulich (1993). In this instrument, respondents

485

were asked to link or match each of eleven nutrients (i.e., sodium, calcium, vitamin A, protein, vitamin C, iron, vitamin D, carbohydrates, saturated fat, potassium and dietary fiber) with an appropriate health consequence from the following list: high blood pressure, strong bones, healthy vision, production of amino acids, fighting colds, delivering oxygen through the bloodstream, calcium absorption, conversion to sugar to fuel the body, cardiovascular disease, balancing sodium, and colon health. An index of health knowledge was constructed by adding all correct answers for each respondent (hence the index ranges from a minimum of 0 representing no health knowledge to a maximum of 11 representing the highest health knowledge). Nutritional awareness refers to consumers’ awareness of the importance of dietary choices in preventing diseases. Once again, well-established multi-item scales do not exist for this construct. Following Churchill (1979), we developed and refined a set of eight measurement items (Cronbach’s alpha = 0.875). Responses to these items were summed up to obtain an index of nutritional behavior/concerns. Table 2 provides brief descriptions and summary statistics for all the items/constructs used to estimate our models. Estimation results and discussion In order to evaluate the role of soy food consumption status in determining behavioral intentions of respondents, the following three groups of respondents were identified: (1) regular users; (2) infrequent users, and (3) non-users. Respondents who were identified as regular users were asked whether they would be willing to increase their soy consumption in response to one of the three treatments (FDA health claim, general health claim, no claim). Similarly, infrequent users were asked whether they would be willing to consume soy-based food regularly, while non-users were asked whether they would be willing to try soy-based food in response to one of the above three treatments. We formally assess the roles of health claims and soy consumption status with two regression approaches as delineated in the previous section: (i) estimating separate regression models in correspondence with each of the regular, infrequent, and non-user groups, and (ii) aggregating data across the three groups of consumers and estimating a combined regression model that includes dummy variables representing the three groups of consumers along with interaction terms between health claims and soy consumption status. Given the ordered and qualitative nature of our dependent measures, the three regression models (1)–(3) were estimated with ordered probit model (McKelvey and Zavonia, 1975). Estimation results of the three ordered probit regression models are presented in Table 3. LR test statistics decisively reject that the hypothesis across the three models that all coefficients are simultaneously zero (p = 0.001), demonstrating the overall significance of the specified models. Scaled R-squares (0.2257; 0.2718; 0.2869) present further evidence that the three models perform reasonably well for cross-sectional data. The models include three health-related variables: (i) health knowledge, (ii) nutritional awareness, and (iii) perceived health benefits of soy foods. Health knowledge had a significant and positive effect only on the behavioral intentions of non-soy users (t = 2.2951), indicating that non-soy users who have more knowledge about the health consequences of various nutrients are likely to exhibit more positive behavioral intention to try soy-based food products. Yet, the non-significance of the variable in the models for Infrequent and Regular users does not suggest its irrelevance in explaining respondents’ behavioral intentions toward soy-based foods. Instead, it is likely to indicate that the effect of general health knowledge is mediated by soy specific health knowledge

486

W. Moon et al. / Food Policy 36 (2011) 480–489

Table 3 Behavioral intentions toward soy foods: estimated parameters of ordered probit models. Non-soy users

Infrequent users

Regular users

Parameter estimates (b1)

Asymptotic t-values

Parameter estimates (b2)

Asymptotic t-values

Parameter estimates (b3)

Asymptotic t-values

Constant Health knowledge Nutrition awareness

4.9098*** 0.0193** 0.0282***

22.940 2.2951 7.1890

5.3827*** 0.0013 0.0244***

13.940 0.0968 3.3468

4.4830*** 0.0327 0.0500***

12.174 1.2568 6.8192

Perceived attributes Health Benefits Taste Convenience Price

0.0904*** 0.1444*** 0.0843*** 0.0444

9.3863 8.4583 3.9626 1.3582

0.0865*** 0.2337*** 0.0239 0.0316

4.7941 7.8434 0.6452 0.6228

0.1146*** 0.1441*** 0.0112 0.0063

6.6131 4.5388 0.3167 0.1389

Socio-demographics Age Gender White Black Asian

0.0067*** 0.0391 0.0006 0.1212 0.0537

3.5991 0.7917 0.0053 0.9070 0.0302

0.0052* 0.0048 0.2300 0.5243** 0.4871**

1.7005 0.0571 1.4103 2.5570 2.5417

0.0044 0.0063 0.0468 0.0901 0.0683

1.4600 0.0728 0.2706 0.4043 0.3675

Stimuli FDA health claim General health Claim Mu1 Mu2 Mu3

0.3592*** 0.4017*** 1.0573*** 2.1428*** 2.8728***

6.0559 6.7144 20.35 36.66 45.38

0.5456*** 0.5794*** 1.2371*** 2.7473*** 3.7529***

5.4391 5.8082 16.72 28.65 28.94

0.0555 0.1655 1.1446*** 2.5793*** 3.3413***

0.5449 1.5127 18.89 24.44 19.13

Number of respondents Log likelihood value LR (zero slopes) test Scaled R-square

1987 2676.12 486.688 [.000] 0.2257

735 867.737 219.573 [.000] 0.2718

731 791.787 231.038 [.000] 0.2869

Note: Behavioral intentions were measured with the following questions: willingness to try soy-based food for non-soy users; willingness to include soy-based food regularly in their diet for infrequent users; and willingness to increase the consumption frequency of soy-based food for regular users. The following five-point scale was used: (1) definitely would not, (2) probably would not, (3) might or might not, (4) probably would, and (5) definitely would. *** p < 0.01. ** p < 0.05. * p < 0.1.

(perceived health benefits) as has been reported by Moon et al. (2005) that the effect of health knowledge on soy consumption behavior is mediated by soy-specific health knowledge. In contrast to the largely indirect effect of health knowledge, nutritional awareness had a strong and direct effect on all three equations, indicating that respondents who exhibit more nutritional awareness are more likely to try soy-based foods (for nonsoy users), to include soy foods in their diet regularly (for infrequent users), and to increase the consumption frequency of soy-based foods (for regular soy users). Four perceived attributes of soy food (health benefits, taste, convenience, and price) were included in the ordered probit models. Respondents’ perceptions about health benefit and taste turned out to be strongly relevant in determining their behavioral intentions toward soy food: i.e., when respondents perceive health benefits from soy-based food or consider soy food tasty, they are more likely to try, use them regularly, and increase the consumption frequency of soy food. Perception about convenience was significant only in the model for non-soy users. This indicates that non-soy users who consider soy food convenient to eat are more likely to show positive intentions to buy soy food when compared to those who do not share this perception of convenience. The perceptions of current soy users (infrequent and regular) about convenience did not influence their behavioral intentions. This result may indicate that once consumers became experienced with soy foods, then the fear about the lack of knowledge on how to cook disappears and the notion of convenience becomes irrelevant. The two stimuli (FDA health claim and general health claims) produced interesting results. In support of our first hypothesis (H1), both types of health claims had a strong effect on the behavioral intentions of non-soy users and infrequent users after controlling the effects of socio-demographic and psychological

variables. Yet, for regular soy users, the two stimuli did not have statistically significant effects on behavioral intentions (as measured with willingness to increase soy consumption frequency). When combined, these results indicate that the two stimuli exerted differential effects between non-soy and infrequent users and regular users. Upon further examining the results of both non-soy users and infrequent users, the impact of FDA statement (b = 0.3592 and t-value = 6.0559 for non-soy users; b = 0.5456 and t-value = 5.4391 for infrequent users) did not appear different from that of simple health benefit statement (b = 0.4017 and t-value = 6.7144 for non-soy users; b = 0.5794 and t-value = 5.8082 for infrequent users). We formally conducted t-tests with the hypothesis that the impact of FDA health claim does not statistically differ from that of general health claim. The null hypothesis was H0: bFDA = bGeneral and t-values were calculated by t = bFDA  bGeneral/ [Var (bFDA) + Var (bgeneral)  2 Cov (bFDA, bgeneral)]. The calculated t-values were 0.389 and 0.226, failing to reject the null hypothesis. Hence, the test results decisively show that the magnitudes of impacts of the two stimuli did not statistically differ. With respect to our research question (RQ1), this result indicates that FDA’s approval of the health benefits of soy food did not add any further information to the respondents beyond the scientific research results. Table 4 presents the results of the aggregate model. While confirming the previous results of the three separate models with regard to the psychological and socio-demographic variables, the aggregate model presents two additional results. First, soy consumption status had a significant effect on behavioral intentions toward soy food: i.e., soy users (infrequent and regular users) are strongly willing to increase their consumption of soy food when compared to non-soy users’ intensity of behavioral intention to use soy foods, supporting the second hypothesis (H2) that

487

W. Moon et al. / Food Policy 36 (2011) 480–489 Table 4 Behavioral intentions toward soy-based food: all respondents.

Effect of FDA Health Claims (βFDA)

All respondents Parameter estimates

Asymptotic t-values

Constant Health knowledge Nutritional awareness

(b) 7.5760 0.0064 0.0313***

30.458 1.0107 10.166

Perceived attributes Health benefits Taste Convenience Price

0.0925*** 0.1553*** 0.0579*** 0.0266

12.311 11.787 3.5830 1.1475

Socio-demographics Age Gender White Black Asian

0.0058*** 0.0252 0.0458 0.1937* 0.1552

4.1495 0.6688 0.5489 1.9958 1.5468

Soy consumption status Infrequent users Regular users

0.2852*** 0.8488***

3.4984 9.4956

Stimuli FDA health claim General health claim

(c) 0.3901*** 0.4163***

6.5915 6.9558

Interaction terms FDA  Infrequent FDA  Regular General  Infrequent General  Regular Mu1 Mu2 Mu3

(h) 0.0433 0.3364*** 0.0622 0.2683** 2.4574*** 3.5775*** 4.7817***

0.3830 2.9074 0.5523 2.3067 13.74 19.86 26.28

Number of respondents Log likelihood value LR (zero slopes) test Scaled R-square

3456 4410.39 1689.04 [.000] 0.4130

Soy Consumption Status Non-soy users

Note: Behavioral intentions were measured with the following questions: willingness to try soy-based food for non-soy users; willingness to include soy-based food regularly in their diet for infrequent users; and willingness to increase the consumption frequency of soy-based food for regular users. The following five-point scale was used: (1) definitely would not, (2) probably would not, (3) might or might not, (4) probably would, and (5) definitely would. * p < 0.1. ** p < 0.05. *** p < 0.01.

behavioral intentions depend upon the degree of engagement in soy food market. With respect to the four interaction terms, note that the coefficients associated with regular soy users were statistically significant and negative (b = 0.3364 and t = 2.9074 for FDA  Regular; b = 0.2683 and t = 2.3067 for General  Regular). These negative signs indicate that the effects of health claims on behavioral intentions are smaller with regular soy users when compared to non-users or infrequent- soy users (refer to Fig. 3 for visual illustration of this result). Consistent with the non-significant effects of the two health claims in the separate model for regular users, the result supports our hypothesis (H3) that soy consumption status moderates the impacts of health claims on behavioral intentions. That is, neither the FDA health claim nor the general health claim instigate regular soy users to be more willing to increase soy consumption. Three explanations are possible for this result. First, the stimuli may not have provided new information to regular users: i.e., they may be already aware of the FDA decision and the health benefits of soy foods. Second, soy users may have determined that their current consumption level is appropriate to benefit from the health-promoting attributes of soy foods. Finally, and perhaps most important, FDA did not emerge as a credible source when evaluating the impact of health claim treatments.

Infrequent users

Regular users

Fig. 3. Smaller impact of FDA health claims with regular soy users.

The ordered probit models (both separate and aggregate) included major socio-demographic profiles such as age, gender, and ethnicity. While gender was not statistically significant, age and ethnicity had a significant association in determining behavioral intentions toward soy food. Age had a negative sign, indicating that older consumers are less likely to exhibit positive behavioral intentions. We might have expected that older consumers would be more willing to consume soy food given the presumption that they would want to take advantage of the health benefits that soy food offers. Yet, recall that consumers’ health knowledge and perceptions of the health benefits of soy food are already controlled in the models. Or, the nature of the relationship between age and behavioral intentions toward soy foods may be nonlinear and could not be captured by the linear models in this article. For example, there could be a negative relationship up to a certain age (say, 50), then the relationship starts to turn positive, indicating that as consumers become older after reaching the age of 50, they tend to show greater willingness to use soy foods. Ethnicity was playing a role only in the model for infrequent soy users: Blacks and Asians were more likely to show positive behavioral intentions when compared to other ethnicity including Whites. Conclusions The American diet has become a vital societal issue that transcends the realms of individual consumers and the food industry. A considerable amount of societal resources is being invested to deal with diet-disease linkages from multiple perspectives (i.e., academics, government, and food businesses). In particular, social/behavioral scientists have responded by researching consumer behavior in relation to dietary choices. Given that soy-based food products have been emerging in recent years as a major functional food, this study investigated whether or not two types of health claims (FDA health claim and general health claim) impact consumers’ behavioral intentions, after controlling for other factors such as soy consumption status, perceived attributes of soy foods, socio-demographic profiles, health knowledge, and nutritional awareness. Building on the finding of Moon et al. (2005) that soy-specific health knowledge significantly impacts consumers’ decisions of whether or not to include soy food in their diets and how often to consume, this study assessed the efficacy of FDA approved health claims as a tool for motivating non-users and infrequent users to get engaged in soy food market. In order to properly identify the effects of health claims on behavioral intentions, we randomly assigned the three stimuli of health claims across respondents. Estimation results from three ordered probit models show that non-soy users and infrequent soy users who were exposed to either FDA health claim or general health claim are significantly

488

W. Moon et al. / Food Policy 36 (2011) 480–489

more likely to eat soy-based food products. FDA or General health claim, however, did not change the behavioral intentions of regular soy users. These results hint that soy consumption status moderates the impacts of health claims on behavioral intentions. While showing that regular soy users are significantly more likely to show positive behavioral intentions toward soy-based food products when compared to infrequent and non-soy users (main effects), the aggregate model explicitly confirmed the moderating roles of soy consumption status on the effects of FDA and general health claims on behavioral intentions. The confirmed role of health claims in this article extends the results of prior research (Ippolito and Mathios, 1991; Ghani and Childs, 1999; Roe et al., 1999; Garretson and Burton, 2000; Kozup and Creyer, 2003) specifically in the context of consumer behavior with respect to soy food. In particular, the significant association of health claims with the behavioral intentions of non-users and infrequent soy users suggests that there is a considerable potential for soy food market to grow and for public health to benefit from the increased consumption of soy food. Several implications emerge for food industry strategies and public health policies. First, health claims are most effective in altering consumer behavior when they are targeted at non-users or infrequent soy users. A question/challenge is how to reach these two groups to convey the FDA approved health claims. Clearly, these two groups are most important because regular soy users are already knowledgeable about soy benefits (and therefore likely engaged in optimal consumption of soy food), so the FDA health claim may not perceptibly change the attitudes or behavior of regular soy consumers. For our society to reap the full benefits of the health-promoting properties of soy proteins, it is crucial for both the food industry and public health agencies to develop information dissemination programs that penetrate non-users and/or infrequent soy users. Second, showcasing the FDA approval of health benefits of soy proteins within a health claim did not stimulate respondents to be more positive toward soy food when compared to a health claim that simply described the scientific clinical research results. The finding suggests that the word ‘‘FDA’’ did not add any further information to consumers beyond the general health claim. This finding is in contrast to previous research showing that consumers attached significant credibility to FDA approval/certification (particularly compared to USDA and no certification) in the case of consumers’ evaluation of genetically modified (GM) food products (Teisl et al., 2008; Roe and Teisl, 2007; Teisl et al., 2003). Such mixed results suggest that there are needs for further research taking an in-depth look at the role of FDA certification in determining consumers’ perceptions, attitudes, behavioral intentions, and behaviors in relation to food products or food technology. In particular, given that this study was specifically in the context of soy-based food products, a question arises as to whether or not consumer evaluation of the credibility of public health-related agencies differs across different products. Hence, future research needs to evaluate the role of FDA certification across different products and assess whether or not consumers’ evaluations of FDA credibility are product-specific. References Ajzen, I., Fishbein, M., 1977. Attitude–behavior relations: a theoretical analysis and review of empirical analysis. Psychological Bulletin 84, 888–918. Ajzen, I., Fishbein, M., 2005. The influence of attitudes on behavior. In: Albarracin, D., Johnson, B.T., Zanna, M.P. (Eds.), The Handbook of Attitudes. Erlbaum, Mahwah, NJ. Anderson, J.W., Johnstone, B.M., Cook-Newell, M.A., 1995. Meta-analysis of the effects of soy protein intake on serum lipids. New England Journal of Medicine 333, 1715–1716. Baron, R.A., Kenny, D.A., 1986. The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology 51, 1173–1182.

Beharrell, B., Denison, T.J., 1995. Involvement in a routine food shopping context. British Food Journal 97, 24–29. Brown, D.J., Schrader, L.F., 1990. Cholesterol information and shell egg consumption. American Journal of Agricultural Economics 72, 548–555. Burton, Scot, Creyer, Elizabeth H., 2004. What consumers don’t know can hurt them: consumer evaluations and disease risk perceptions of restaurant menu items. Journal of Consumer Affairs 38, 121–145. Calfee, John E., Pappalardo, Janis K., 1991. Public policy issues in health claims for foods. Journal of Public Policy & Marketing 10, 33–53. Caswell, J.A., Padberg, D.I., 1992. Toward a more comprehensive theory of food labels. American Journal of Agricultural Economics 74, 460–468. Caswell, Julie A., Ning, Yumei, Liu, Fang, Mojduszka, Eliza M., 2003. The impact of new labeling regulations on the use of voluntary nutrient-content and health claims by food manufacturers. Journal of Public Policy & Marketing 22, 147– 158. Chern, W.S., 2002. Health, nutrition and demand for food: an American perspective. In: Chern, W.S., Rickersten, K. (Eds.), Health, Nutrition and Food Demand. CABI Publishing, Cambridge, MA. Chern, W.S., Loehman, E.T., Yen, S.T., 1995. Information, health risk beliefs, and the demand for fats and oils. Review of Economics and Statistics 77, 555–564. Churchill, G.A., 1979. A paradigm for developing better measures of marketing constructs. Journal of Marketing Research 16, 64–73. Devine, D., 2002. Soya and health 2002 – clinical evidence, dietetic applications. British Nutrition Foundation Nutrition Bulletin 27, 1195–1198. FDA, 1998. Notification of a health claim or nutrient content claim based on an authoritative statement of a scientific body. Center for Food Safety and Applied Nutrition, Office of Food Labeling, (accessed 7.12.08). Ford, Gary T., Hastak, Manoj, Mitra, Anusree, Ringold, Debra Jones, 1996. Can consumers interpret nutrition information in the presence of a health claim? A laboratory investigation. Journal of Public Policy & Marketing 15, 16–27. France, K.R., Bone, R.F., 2005. Policy makers’ paradigms and evidence from consumer interpretations of dietary supplement labels. Journal of Consumer Affairs 39, 27–51. Freedonia Group Inc., 2005. Soy Products & Markets, Product Code: R154-1359. Garretson, J.A., Burton, S., 2000. Effects of nutrition facts panel values, nutrition claims, and health claims on consumer attitudes, perceptions of disease-related risks, and trust. Journal of Public Policy & Marketing 19, 213–227. Ghani, W.I., Childs, N.M., 1999. Wealth effects of the passage of the nutrition labeling and education act of 1990 for Large US. Journal of Public Policy & Marketing. Goland, E., Unnevehr, L., 2008. Food product composition, consumer health, and public policy: introduction and overview of special section. Food Policy 33, 465–467. Gordon, M.E., McKeage, K., Fox, M.A., 1998. Relationship marketing effectiveness: the role of involvement. Psychology and Marketing 15, 443–459. Hasler, C.M., 2002. The cardiovascular effects of soy products. Journal of Cardiovascular Nutrition 16, 50–63. Ippolito, P.M., Mathios, A.D., 1991. Health claims in food marketing: evidence on knowledge and behavior in the cereal market. Journal of Public Policy & Marketing 10, 15–32. Ippolito, Pauline M., Mathios, A.D., 1994. Information, policy, and the sources of fat and cholesterol in the US diet. Journal of Public Policy & Marketing 13, 200–217. Ippolito, P.M., Mathios, A.D., 1995. Information and advertising – the case of fat consumption in the United States. American Economic Review 85, 91–95. Jain, S.P., Posavac, Steven S., 2001. Prepurchase attribute verifiability, source credibility, and persuasion. Journal of Consumer Psychology 11, 169–180. Keller, Scott B., Landry, Mike, Olson, Jeanne, Velliquette, Anne M., Burton, Scot, Andrews, J. Craig., 1997. The effects of nutrition package claims, nutrition facts panels, and motivation to process nutrition information on consumer product evaluations. Journal of Public Policy & Marketing 16, 256–269. Kozup, J.C., Creyer, E.H., 2003. Burton, scot, making healthful food choices: the influence of health claims and nutrition information on consumers’ evaluations of packaged food products and restaurant menu items. Journal of Marketing 67, 19–34. Mathios, A.D., 1998. The importance of nutrition labeling and health claim regulation on product choice: an analysis of the cooking oils market. Agricultural and Resource Economics Review 27, 159–168. McKelvey, R., Zavonia, W., 1975. A statistical model for the analysis of ordinal level dependent variables. Journal of Mathematical Sociology 4, 103–120. Messina, M., Barnes, S., 1995. The role of soy products in reducing risk of cancer. Journal of National Cancer Institute 83, 541–546. Mojduszka, E.M., Caswell, J.A., 2000. A test of nutritional quality signaling in food markets prior to implementation of mandatory labeling. American Journal of Agricultural Economics 82, 298–309. Moon, W., Balasubramanian, S.K., Rimal, A., 2005. Perceived health benefits and soy consumption behavior: two-stage decision model approach. Journal of Agricultural and Resource Economics 30, 315–332. Moorman, C., Matulich, E., 1993. A model of consumers’ preventive health behaviors: the role of health motivation and health ability. Journal of Consumer Research 20, 208–228. Nayga Jr., R.M., Lipinski, D., Savur, N., 1998. Consumer’s use of nutritional labels while food shopping and at home. Journal of Consumer Affairs 32, 4. Pszczola, Donald, 2000. Soy: why it’s moving into the mainstream. Food Technology 54, 76–84.

W. Moon et al. / Food Policy 36 (2011) 480–489 Roe, B., Teisl, M.F., 2007. Genetically modified food labeling: the impacts of message and messenger on consumer perceptions of label credibility and adequacy and of product risk. Food Policy 32, 49–66. Roe, B., Levy, A.S., Derby, B.M., 1999. The impact of health claims on consumer search and product evaluation outcomes: results from FDA experimental data. Journal of Public Policy and Marketing 18, 89–105. Roosen, J., Marette, S., Blanchemanche, S., Verger, P., 2009. Does health information matter for modifying consumption? A field experiment measuring the impact of risk information on fish consumption. Review of Agricultural Economics 31, 2– 20. Soyatech Inc. and Spins Inc. Soyfoods, 2005. The US Market, Fifth Annual Edition. Sternthal, Brian, Phillips, Lynn W., Dholakia, Ruby, 1978. The persuasive effect of source credibility: a situational analysis. Public Opinion Quarterly 42, 285–314. Teisl, M.F., Garner, L., Roe, B., Vayda, M., 2003. Labeling genetically modified foods: how do US consumers want to see it done? AgBioForum 6, 48–54. Teisl, M.F., Radas, S., Roe, B., 2008. Struggles in optimal labeling: how different consumers react to various labels for genetically modified foods. International Journal of Consumer Studies 32, 447–456. Unnevehr, L.J., Hasler, C., 2000. Health claims and labeling regulation: how will consumers learn about functional foods. AgBioForum 3, 10–13.

489

Variyam, J.N., Blaylock, J., Smallwood, J., 1998. Informational effects of nutrient intake determinants on cholesterol intake. Journal of Agricultural and Resource Economics 23, 110–125. Verbeke, W., Vackier, I., 2004. Profile and effects of consumer involvement in fresh meat. Meat Science 67, 159–168. Wansink, B., 2003. How do front and back package labels influence beliefs about health claims. Journal of Consumer Affairs 37, 305–316. Wansink, B., Chan, N., 2001. Relation of soy consumption to nutritional knowledge. Journal of Medicinal Food 4, 145–150. Wansink, B., Cheney, M.M., 2005. Leveraging FDA health claims. Journal of Consumer Affairs 39, 386–398. Wansink, B., Park, S.B., Sonka, B.S., Morganosky, M., 2000. How soy labeling influences preference and taste. International Food and Agribusiness Management Review 3, 85–94. Wansink, B., Westgren, R.E., Cheney, M.M., 2005. Hierarchy of nutritional knowledge that relates to the consumption of a functional food. Nutrition 21, 264–268. Wiener, Joshua L., Mowen, John C., 1986. Source credibility: on the independent effects of trust and expertise. Advances in Consumer Research 13, 306–310. Zaichkowsky, J.L., 1985. Measuring the involvement construct. Journal of Consumer Research 12, 341.