Additives are used by the food industry to extend the shelf life of food, enhance consumer appeal, and ensure the safety and quality of food [1
]. However, consumers often look at food additives with suspicion and try to avoid them [2
]. Most consumers have little knowledge of food additives and perceive them as harmful to their health [3
]. Consumer concerns over additives may affect food manufacturers who process food as well as policymakers who endeavor to communicate effectively on food safety issues.
A notable example of consumer mistrust and concern over food additives is the monosodium glutamate (MSG) issue in South Korea. This is the most longstanding and frequently discussed issue relating to food safety in the country, with the majority of consumers believing that the additive is harmful [4
]. The MSG issue has continued for decades in the country because of continued assertions of its hazards by mass media and civil groups.
The controversy over MSG first arose in Western countries. MSG is the salt form of glutamic acid, and it is used as a food additive because of its flavor-enhancing properties. Some people believe it causes a number of allergy-like symptoms, including but not limited to the following: MSG-induced asthma; migraine headaches; hives; swelling of the face, mouth, and tongue; nausea; and rhinitis. However, most of the evidence comes from anecdotal cases and dubious clinical studies. Decades of research has failed to demonstrate a relationship between MSG consumption and the development of serious reactions in most people. In 1987, the U.S. Food and Drug Administration (FDA) placed MSG in the category of “generally recognized as safe” (GRAS) [6
], based on its extensive history of use in food before 1958 and on published scientific evidence. Other examples of GRAS substances include salt, sugar, spices, and vitamins [7
]. The Joint Expert FAO/WHO Committee on Food Additives [8
] classified MSG as an additive with an “ADI (acceptable daily intake) not specified,” indicating that no toxicological concerns arise from its use as a food additive.
The dose-dependent effects of food additives in humans are evaluated by the ADI and estimated daily intake (EDI). The maximum allowable dietary exposure to a food additive is related to ADI, which is based on effects of food additives observed in animal models; specifically, the no-observed-adverse-effect level (NOAEL) determined in animal studies from their lifetime exposure to the substance. EDI is an indication of cumulative exposure to a given substance and reflects the concentration of that substance in the food itself as well as the amount of consumption of the food. Both ADI and EDI are determined for each food additive. U.S. FDA approval of a food additive is contingent upon the stipulation that the sum of all EDIs of a food additive from all sources must not exceed the ADI. The ADI for MSG is 120 mg/kg/d [9
Although the controversy over MSG seems to have abated in Western countries, it is still a matter of concern in South Korea. The Korea Food and Drug Administration [10
] has accepted MSG as safe and has allowed its unlimited use in South Korea. According to Sah and Yeo [11
] and Woo [12
], Korean consumers with more limited knowledge of food additives tend to be more concerned because they do not understand scientific facts related to safety issues.
This study investigated two research questions regarding the MSG issue in South Korea. The first issue covers knowledge cultivation effects on consumers’ responses to the additive issue. Previous studies have suggested that excessive concerns over controversial food safety issues can be mitigated by developing more reasonable risk perceptions based on enhanced consumer knowledge [13
]. There may be several important mediators concerning the enhancement of consumer knowledge regarding food safety and hazards, such as media information, consumer education, or government publicity. In the present study, we focus on education opportunities. Positive effects of risk education on recipients’ knowledge of health risks and desirable risk behaviors have been suggested in the literature [18
]. The second issue is related to the message frame of food safety news. News presented by mass media and from scientists’ opinions is important in food safety issues [22
]. Separate from the cultivation of consumer knowledge, specific frames used in media coverage can play a significant role in responses by the public [24
]. This suggests that consumer responses, and thus purchasing behaviors, may differ according to the framing of information conveyed by the media. Radley [25
], Dorfman [26
], Cohen et al. [27
], and Jin and Han [13
] showed that consumers’ cognition and responses are affected by message frames. In addition, food safety knowledge and message framing may interact with each other. Jin and Han [13
] revealed that people with less knowledge about a food hazard will have higher levels of disturbance due to reports conveyed by mass media regarding a particular food safety issue.
For empirical application of the research questions, we used data from a survey of college students. First, we investigated whether students’ experience in a food safety class affected their responses to the MSG issue. The content of the class included food hazards, scientific facts about food additives, and the role of food-related institutions, including the government. Students’ class experience could have resulted in a higher level of knowledge of MSG through enhanced understanding of scientific facts related to safety issues and/or increased personal interest in food additives, although the class did not specifically cover the MSG issue. Accordingly, we hypothesized that students who attended the food safety class would have relatively better knowledge of MSG than those who did not, making them less fearful of the additive. We used a structural equation model (SEM) to examine the relation process.
Second, we analyzed students’ responses to message framing as well as the interaction between class experience and message framing. That is, we tested whether a message frame with different information constitutions changed students’ intention to avoid MSG and whether students who took the class were affected less by message framing than those who did not attend the class. Jin and Han [13
] showed that message framing regarding food safety issues has an influence on college students’ responses. Participants of the study revealed divergent purchase intentions in response to different headlines and different amounts of information within articles. Those who had less knowledge had more variation in their purchase intentions in response to different message frames. Here, we implemented different message frames regarding the MSG issue. We provided a short article with both positive and negative information regarding MSG to half of the survey respondents and another article with only positive information to the other half, producing a 2 × 2 factorial design, with class experience. The hypotheses tested were that different message frames cause different responses by the students and that the group who had taken the class would be less affected by the message framing.
The remainder of this paper is organized as follows: Section 2
presents the analytical methods and data; Section 3
presents the estimation results and discussions; and Section 4
concludes the paper.
3. Results and Discussion
3.1. Results of SEM Estimation
We conducted a confirmatory factor analysis (CFA) before estimating the SEM model to assess the construct validity and reliability of the measurement model. Further, we checked the construct validity by examining the convergent validity and discriminant validity. The results indicated statistical significance of the t-values associated with standardized loadings (p
< 0.01), suggesting a sufficient degree of convergent validity. The results also confirmed that no squared correlation exceeded the extracted average variance, indicating that the criterion for discriminant validity was met. In addition, all the Cronbach’s alpha and composite reliability values were above 0.7, indicating acceptable reliability [48
]. Overall, the reliability and validity analysis results indicated satisfactory measurement qualities. The model fit statistics showed that χ2
/df was 3.35 (which is close to 3), the normed fit index (NFI) was 0.795, the comparative fit index (CFI) was 0.837, the Akaike information criteria (AIC) was 64.83, and the root mean squared error of approximation (RMSEA) was 0.091 (which is close to zero). The results suggested a reasonable goodness-of-fit for the model [49
]. The estimation results are shown in Table 4
The SEM results showed that students with more knowledge of MSG demonstrated lower levels of risk perception (supporting H1). However, students’ social trust did not reveal any influence on risk perception (not supporting H2). Attitude negatively affected risk perception (supporting H3). Specifically, it showed that if students had more negative attitudes toward food additives and foods in safety controversies, they perceived MSG as riskier. Int-avoid was influenced by risk perception (supporting H4), suggesting that students are more likely to avoid MSG when their risk perception of MSG is relatively higher. We further found that class experience increased students’ objective knowledge of MSG (supporting H5). Class experience also had a positive impact on participants’ social trust in food safety and related institutions and groups (supporting H6). However, class experience did not show a statistically significant influence on the students’ attitudes toward food additives and foods in safety controversies (not supporting H7).
We further derived the indirect effect of class experience on risk perception and int-avoid. Using the bootstrap method, we derived the standard error of the indirect effect and therefore, the statistical significance of each effect. The findings in Table 5
show that students with the class experience showed less risk perception regarding MSG and less intention to avoid MSG.
3.2. Results of Ordered Logit Estimation
The results of the ordered logistic regression are presented in Table 6
. Since there was a relatively large number of explanatory variables, the multicollinearity among the variables was checked using the variance inflation factor (VIF). All estimated VIFs were less than 2.0, meaning that multicollinearity was not a significant concern in the estimation.
As presented in the table, there are three different models: (1) a model with only the variables of class experience, message frame, and class experience × message frame, (2) a model in which demographic variables were added to Model 1, and (3) a model in which social trust, knowledge, attitude, and perceived risk were added to Model 2. Class experience and message frame were statistically significant throughout all models. Class experience worked to decrease the int-avoid value. Message frame worked to increase the int-avoid rate, meaning that when students were provided with both positive and negative information regarding MSG, their intention to avoid MSG increased compared to the case in which only positive information was provided. This means that the message frame had an influence on the students’ responses as measured by purchase intention.
The coefficient of class experience × message frame, , had a negative value, which means that the observed int-avoid value of the treatment group, when faced with frame B, is smaller than the value derived by the assumed counterfactual. That is, the responses to the message frame by students who took the class were weaker than those from students who did not take the class. However, the coefficient was not statistically significant at any conventional significance level throughout the models, which means that different responses to the message frame, according to different status of class experience, were not statistically supported. This may be because the college students were sensitive to the message framing itself, regardless of their class experience status, or because we provided a strong form of message framing: positive information versus positive and negative information together. If we used other message frames such as different information details or different nuance in headlines, we might have obtained statistically significant support for different responses to message frames based on the class experience.
Information and education would be important mediators in enhancing consumer knowledge regarding MSG. An educational effect on the MSG issue has not been found in the literature. However, we did find studies analyzing the role of MSG information in shaping consumer responses, and the findings of those studies are inconclusive. Wang and Adhikari [50
] performed a cross-sectional survey examining U.S. consumers’ perceptions of MSG and found that consumers who have information about MSG are more likely to report higher risk perception of MSG than their counterparts. Conversely, Greenacre et al. [51
], using an in-depth online panel survey through Amazon’s Mechanical Turk, found that factual information in the shape of a rational appeal played a crucial role in increasing the likelihood of consuming MSG, suggesting the importance of providing accurate health information based on the true nature of MSG. A key difference between these two studies is the nature of the information respondents have or were provided; that is, subjective/emotional versus objective/rational. This implies that both the way information is conveyed as well as what information is conveyed are important, which is related to our second research question.
Our second result indicated the sensitivity of students’ responses to message framing, although sensitivity did not differ statistically by whether students had the class experience. This confirms that the delivering frame of food safety news is important, suggesting that specific frames used in media coverage can play a significant role in responses by the public. This implies that consumer responses, and thus purchasing behaviors, may differ according to how and/or what information is conveyed by the media. The findings have implications to which the government, mass media, and food makers need to pay attention when providing news or information regarding food hazard and safety. Additionally, the food industry and policymakers need to communicate with consumers with objective and correct information, and they need to educate consumers on a factual basis regarding MSG.
The results of our first research question are aligned with those of Greenacre et al. [51
], given that opportunity to participate in a food safety class provides recipients with more knowledge of food hazards and safety and thus make their perceptions of controversial food safety issues more reasonable. Greenacre et al. [51
] imply that food manufacturers need to place correct information about the true nature of MSG in their advertising and on their food products rather than allowing incorrect beliefs to pervade. This finding, together with our results, imply that in order to reduce unnecessary consumer concern, consumers’ knowledge of food additives and scientific facts need to increase through education on a factual basis. Shim et al. [52
] reported that 76.8% of South Korean respondents felt information on food additives is insufficient. They attributed this lack of information to difficulties in understanding the subject of food additives as well as insufficient education and public relations campaigns. Therefore, our results suggest recommendations for regulatory agencies and food-related industries—they need to develop an educational curriculum related to food additives and scientific facts, in addition to transmitting information regarding additives to the public. If governments, schools, or other relevant institutions or groups could provide the public with opportunities to participate in food safety education, recipients would have a better understanding of such issues, and it would result in more realistic risk perceptions and rational responses to food safety issues.
We analyzed whether students’ experience in a food safety class would result in more realistic responses to the controversial MSG issue. Our findings suggested that although students with class experience did not specifically discuss the MSG issue, they had relatively more knowledge of MSG and a lower risk perception or fear of MSG, resulting in a reduced intention to avoid MSG. Their class experience also increased their trust in overall food safety in the domestic market as well as in food-related institutions and groups. Students showed sensitive responses to message framing, but the sensitivity did not statistically differ by whether students had the class experience. This may be because college students were sensitive to the message frames regardless of their class experience status or because we provided a strong form of message frame.
This study has several limitations, which lead us to provide certain suggestions for future studies. First, because the survey data were restricted to college students, discretion is required before generalizing the findings to other groups. Second, the findings we derived relate to only one country. Future analyses should examine various countries to arrive at a strong conclusion. Several important control variables, such as personal hazard experience, optimistic bias, and other demographic characteristics, may interact and raise consumers’ knowledge of food additives and desirable risk behaviors, in addition to education. Therefore, future studies should focus on these other factors as well.