Public risk perception of food additives and food scares. The case in Suzhou, China

Appetite 70 (2013) 90–98

Contents lists available at SciVerse ScienceDirect

Appetite journal homepage: www.elsevier.com/locate/appet

Research report

Public risk perception of food additives and food scares. The case in Suzhou, China q Linhai Wu a,b,⇑, Yingqi Zhong a, Lijie Shan a,b, Wei Qin c a

School of Business, Jiangnan University, Wuxi, Jiangsu 214122, China Food Safety Research Base of Jiangsu Province, Jiangnan University, Wuxi, Jiangsu 214122, China c Independent Contributor, 16 York Road, West Windsor, NJ 08550, United States b

a r t i c l e

i n f o

Article history: Received 4 July 2012 Received in revised form 18 June 2013 Accepted 22 June 2013 Available online 4 July 2013 Keywords: Food additives Risk perception Food scares Purchase intention Structural equation modeling (SEM)

a b s t r a c t This study examined the factors affecting public risk perception of food additive safety and possible resulting food scares using a survey conducted in Suzhou, Jiangsu Province, China. The model was proposed based on literature relating to the role of risk perception and information perception of public purchase intention under food scares. Structural equation modeling (SEM) was used for data analysis. The results showed that attitude towards behavior, subjective norm and information perception exerted moderate to high effect on food scares, and the effects were also mediated by risk perceptions of additive safety. Significant covariance was observed between attitudes toward behavior, subjective norm and information perception. Establishing an effective mechanism of food safety risk communication, releasing information of government supervision on food safety in a timely manner, curbing misleading media reports on public food safety risk, and enhancing public knowledge of the food additives are key to the development and implementation of food safety risk management policies by the Chinese government. Ó 2013 Elsevier Ltd. All rights reserved.

Introduction Food safety is a worldwide problem. Food safety risk in China has become more of a direct/indirect result of social behavior than a direct result of natural factors; essentially, it is a social risk associated with human factors (Li, 2011). For various reasons, food safety incidents caused by the abuse of food additives1 continue to occur in China. It has become the most common type of food safety incident and a major public concern (Li, Liu, Wang, & Dai, 2011; Ouyang, 2011). Hence, it is critical that the Chinese government identify ways to mitigate potential food scares resulting from the abuse of food additives. Previous studies demonstrated that purchase intentions were affected by different levels of risk perception when there was a food scare (Mazzocchi, Lobb, Bruce Traill, & Cavicchi, 2008). It should be noted that multiple definitions of food scares exist in the literature. In general, food scares can be interpreted as q

Acknowledgments: This research was supported by the key projects of the Social Sciences for bids from the Colleges of Jiangsu Province, China (No. 2011ZDAXM018), the Natural Science Foundation of Jiangsu Province of China (No. BK2012126), Social Sciences of the Ministry of Education of China (No. 11YJC630172) and Research on Chinese Food Safety Risk Management, Supported by the Central University Basic Research Funds (No. JUSRP51325A). ⇑ Corresponding author at: School of Business, Jiangnan University, Wuxi, Jiangsu 214122, China. 1 Abuse of food additives is the use of food additives beyond the specified amount or range, or use of fake and shoddy or expired food additives.

0195-6663/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.appet.2013.06.091

increasing public anxiety over a continuous increase in food safety incidents, and this anxiety is closely related to the intensity of media attention to such incidences (Knowles, Moody, & McEachern, 2007). In essence, food scares are an external manifestation of public mental activities. In recent years, frequent outbreaks of food safety incidents due to the abuse of food additives in China have attracted close attention of the mass media, and have affected the public’s risk perception of food additive safety, which may in turn lead to food scares. Hence, it is important to focus on the case of food additives to investigate public risk perception of food additives and intention to purchase food containing additives under food scares. As a result, coping strategies could be developed to preserve social stability.

Research framework and hypotheses development Attitudes towards behavior and subjective norm Fishbein’s multi-attribute model assumed that a person’s attitude towards an object is determined by the sum of beliefs that the person has about the consequences or attributes of the object weighted by how they are evaluated (Fishbein, 1963). This model has been widely applied in consumer research. The following reviewed the factors that affected consumer attributes and behavior towards food or food safety issues.

L. Wu et al. / Appetite 70 (2013) 90–98

Theory of Reasoned Action (Ajzen & Fishbein, 1980) pointed out that consumer attitudes are not only based on the consequences that are perceived by the person him- or herself, but also on a person’s belief that significant others think he or she should engage in this behavior. Thus, in empirical research on public’s attitude and purchase intention towards food containing additives, subjective norm, which represent perceptions of significant others’ preferences about whether one should engage in this behavior should be added in the present study. Past studies in various areas, such as attitude towards genetically modified foods (Cook, Kerr, & Moore, 2002), and attitude towards vegetable consumption after education intervention (Kothe, Mullan, & Butow, 2012), have shown that attitudes towards behavior affected food choices. Zagata (2012) analyzed the Czech Republic consumer’s behavioral intentions towards organic food, and concluded that attitude toward the behavior and subjective norm were both good predictors and had a positive influence on consumer’s behavioral intention. In consumer buying behavior after food safety accidents, Mazzocchi et al. (2008) studied the intention to purchase chicken among 2725 consumers in France, Germany, Italy, the Netherlands, and the United Kingdom, and found that attitudes towards behavior, subjective norm and perceived behavior control affected the changes in consumer buying behavior after a food safety accident. Risk perception Risk perception was included in studies analyzing consumers’ purchase intention after food scares. For example, Lobb, Mazzocchi, and Traill (2006) analyzed the chicken buying behavior and risk perception of consumers after the outbreak of avian flu by integrating risk perception and trust into the Theory of Planned Behavior (TPB) framework and considering the influence of different individual (or household) characteristics. Risk perception was also included in studies on genetically modified food. Chen and Li (2007) pointed out that risk perception, benefit perception, knowledge, and trust were important factors affecting the attitudes of Taiwanese consumers toward genetically modified foods. In an analysis of Italian consumers’ intention to purchase genetically modified food, Prati, Pietrantoni, and Zani (2012) introduced risk perception, benefit perception, and trust in government institutions. Qin and Brown (2008) examined consumer attitudes towards genetically engineered salmon, and the results showed that attitudes towards genetically engineered salmon were influenced by risk perception, trust, knowledge, and outrage factors. In recent years, due to the repeated outbreaks of various food safety incidents in China, public confidence in domestic food has been declining. As such, food safety is ranked number one among issues attracting the public attention in China in 2012 (E, 2012). The abuse of food additives and illegal use of chemical additives have become the primary sources of food safety incidents in China (Wu, Zhang, Shan, & Chen, 2012). Because of this, the public risk perception of food additives has grown increasingly strong, thus resulting in scares. Therefore, risk perception, an important factor affecting public food scares, should exert an impact on the intention to purchase food containing additives under food scares; and is included in this context.

91

of food additives in South Korea also found that 76.8% of the participants expressed the view that information on food additives was insufficient, and the participants also claimed that it was difficult to understand the subject of food additives. A similar situation also exists in China. Information that is spread in diversified channels, like a complex crisscrossing network, can easily mislead. Some news media outlets/reporters do not have food safety expertise, and may present misleading information during the dissemination process. Hence, inaccurate or false information is amplified through poor journalism and uncensored social media in China. This misinformation can spread rapidly across the country and easily led to public food safety scares when the public has low scientific literacy and is biased against food additives (Wu & Huang, 2012). Therefore, the public perception of information about food additives has an important influence on public risk perception and purchase intention and should be included in this context. In summary, we propose buying intention of food with additives would be affected by attitude toward the behavior, subjective norm, information perception, and risk perception. Hypotheses Attitudes toward the behavior (ATTI) Affect and cognition have long been considered to be distinct components of attitude (McGuire, 1969). Ajzen (2000) suggested that attitude comprised two specific subcomponents. These were hypothesized to be composed of affective (e.g. enjoyable/unenjoyable) and instrumental (e.g., beneficial/harmful) evaluations toward a behavior. Meiselman and MacFie (1996) introduced negative affect into the TPB to investigate the intention of 172 mothers of children aged 5–11 years in the United Kingdom to choose nutritious foods and food containing additives. They demonstrated that negative affect had a significant impact on the mothers’ choice of nutritious foods and food containing additives. Therefore, it is hypothesized that: H1: Attitudes toward the behavior have an impact on the public risk perception of food additive safety. H2: Attitudes toward the behavior have an impact on public’s purchase intention. Subjective norm (SN) Fu and Tong (2003) suggested that family and reference groups could affect the perception and behavior of the respondents through various information dissemination channels. Sharlin (1987) argued that the exaggerated media reports of food safety incidents might incite an extreme emotional response among the public. In contrast to the organizations or individuals providing positive information of food safety, those providing negative information were more acceptable to the public; therefore, mass media had a greater incentive to provide negative information (Verbeke & Ward, 2001). Public confidence in food safety has always been significantly decreased by negative information, even without scientific evidence (Verbeke & Kenhove, 2002). A study by Wang (2010) of 382 consumers in Taiwan confirmed that consumers’ intention to purchase popular food online was affected by food brand reputation, and that brand reputation had a positive impact on consumer confidence and reduced consumer risk perception. Therefore, it is hypothesized that:

Information The present study investigated the public risk perception of food additives and food scares in Suzhou, China. It should be noted that, many non-edible chemicals maybe misconstrued as food additives due to the absence of public education on this topic, lack of general knowledge and misconception of food additives. Shim et al. (2011) studied consumers’ knowledge and safety perceptions

H3: SN has an impact on the public risk perception of food additive safety. H4: SN has an impact on public’s purchase intention. Information perception (INMF) The public’s perception and evaluation of the safety of additives is related to the information regarding food additives they already

92

L. Wu et al. / Appetite 70 (2013) 90–98

possess, as well as the perception of this information. Stern, Haas, and Meixner (2009) performed a survey on wood-based food additives among 263 Australian respondents. They found that training on basic knowledge regarding wood-based food additives had an active and effective influence on consumers’ attitudes toward purchase. Training also improved consumers’ positive assessment of wood-based food additives. Worsley, Worsley, and McConnon (1991), Worsley (1996) and Williams, Stirling, and Keynes (2004) suggested that information regarding food additives that consumers needed the most was the food label information, and that most consumers determined whether the food contained additives according to the food label, which affected their decision to purchase. Slovic (1993) found that consumers tended to follow expert opinion, which affected consumers’ reaction to risk. Moschitz and Stolze (2010) pointed out that government enhancement of food safety, consisting of communication of food quality information to consumers by labeling the food with a quality and safety certification mark had become an important policy tool. Food quality and safety certification has a direct impact on consumers’ purchase behavior (Falguera, Aliguer, & Falguera, 2012). Therefore, it is hypothesized that:

Methods Sample selection When public safety incidents caused by the abuse of food additives erupt, public risk perception regarding food additive safety is closely related to the public knowledge of food additives. It has been demonstrated that the overall knowledge of food additive is higher in cities with a higher level of economic and the more highly developed society (Liu & Zhang, 2004). Since Suzhou City of Jiangsu Province is one of the highly developed cities in China, we hypothesized that the public’s awareness of food safety in Suzhou, including food additive safety, would be relatively strong, and their understanding of food additive would be more comprehensive, thus we conducted this survey in Suzhou. A pilot survey was conducted to examine the construct validity. Modifications were made and the final survey was administered among residents 18 years of age or older in Suzhou in the form of direct, one-on-one interviews for this study. A stratified sample was draw from the Suzhou population according to its census data. Questionnaire design

H5: Information perception has an impact on the public risk perception of food additive safety. H6: Information perception has an impact on public’s purchase intention. Risk perception (RISK) Fischer, De Jong, De Jonge, Frewer, and Nauta (2005) suggested that the public perception of the risk of food hazards would directly affect their responses to food safety incidents. For example, the more intense the public perceptions of hazards of food safety incidents, the more negative their evaluation of food safety, and the higher the anxiety levels experienced. Verbeke, Frewer, Scholderer, and De Brabander (2007) believed that when the public found it difficult to prevent or eliminate a food safety problem by individual effort, their fear would be enhanced. Furthermore, unfamiliarity, credibility, and unpredictability increase the public’s expectation of scares of food safety crises that may occur at various degrees; thus these factors may also affect their intention to purchase a particular food. Therefore, it is hypothesized that: H7: Risk perception has an impact on the public’s purchase intention. Covariance Perceived risk associated with food with additives is the result of the public confidence in food safety, while public confidence of the food with additives has a great influence on their purchase intention. The greater the risk public perceives of food additives, the more likely they are to lose confidence, and thus, the lower intention of them to purchase. On the other hand, public’s confidence was derived from their information perception, which was affected by the opinions of the people that they trust. Therefore, SN, attitude towards behavior and information may affect consumers’ purchase intention by risk perception, thus risk perception would mediate SN, attitude towards behavior and information to buying intention. Lobb, Mazzocchi, and Traill (2007) also found a significant relationship between consumers’ trust in various information sources, risk perception, and attitudes towards the behavior. Therefore, it is hypothesized that: H8: The public’s attitudes toward the behavior, subjective norm, information perception, have a significant relationship among each other.

In order to ensure content validity of the questionnaire and to verify the hypothetical model shown in Fig. 1, twenty two measurement variables (Table 1) were selected for explanatory variables. In order to facilitate public understanding of the research topic, preservatives in milk were selected as the survey subject to examine the public’s risk perception regarding food additive safety and their purchase intention under food scares. Data collection and missing data analysis In total, 220 questionnaires were distributed during December 14–16, 2011, and 209 valid questionnaires were returned, with a questionnaire response rate of 95%. Results were analyzed using SPSS18.0 software. Eight missing values were identified. Because the missing data were missing at random, they were processed by multiple imputation (MI) in Mplus6.0 (Graham, 2009; Graham, Olchowski, & Gilreath, 2007; Schafer & Graham, 2002). Based on the program results, the complete dataset was obtained after imputation, and was utilized for analysis. Model testing of respondents The statistical results revealed that 80.7% of the respondents were between the age of 20–60 years, 68.3% had attended a junior

Fig. 1. The hypothetical model of factors that influence public risk perception of food additives and the resulting food scares.

L. Wu et al. / Appetite 70 (2013) 90–98

93

Table 1 Variables in the hypothetical model. Latent variable dimension

Code

Measurable variable dimension

Attitude toward the Behavior (ATTI)

FACI SYNT EXCE

Buying food with additives makes me feel unpleasant even though it facilitates our lives Buying food with additives makes me uncomfortable due to the possibility of synthetic additives harming our health Buying food with additives makes me worried because excessive addition of additives is harmful

Subjective Norm (SN)

MEDI FRID FAMI REPU GOVE

Influence on buying food with additives from media reports about the public’s behavior Influence on buying food with additives from family members or friends Unwilling to buy food with additives due to the experience of family Influence on buying food with additives from government policies Influence on buying food with additives from enterprise reputation

Information perception (INFM)

INEN INMD INPR BZXI INSY

Influence from information regarding the presence or absence of quality and safety certification of food with additives Influence from information of media reports regarding food with additives Influence from information regarding experts’ opinions on food with additives Influence from information regarding additives on the package labels Influence from the information of synthetic food additive

Risk perception (RISK)

LCFI MORL HARM RGHC ISLF

Doubt about the food safety of domestic food market Absence of a code of ethics in enterprises makes you suspect the safety of the food with additives Abuse of food additive became a major potential food risk Lose confidence in the domestic food market Government’s regulation on additive is lack of effectiveness

Buy intention (INTEN)

MORE NEVE WONT STAN

Would rather buy expensive food without additives than buy less expensive food with additives Will not buy food with additives under any circumstances Will not buy food with additives Will buy food with additives only if the content of the additives strictly follows the national standards

Table 2 Descriptive statistics of food scares.

Seriousness of the resulta Emotional reactionsb Possibility of Petitionc Confidence in food marketd

1 Not at 2 all (%) (%)

3 (%)

4 (%)

5 (%)

6 (%)

7 Very much (%)

1.4 1.0 5.2 8.6

2.4 2.9 15.8 30.1

2.9 7.7 17.7 23.9

21.0 23.6 20.6 12.0

22.5 33.0 23.0 5.3

49.3 30.6 12.4 1.7

0.5 1.2 5.3 18.4

Note: N = 209. a 1 = not serious at all. . .7 = very serious. b 1 = not angry at all. . .7 = very angry. c 1 = definitely impossible. . .7 = very possible. d 1 = very unconfident. . .7 = very confident.

college or higher, 44.7% were female, the majority (approximately 80%) had a family size of 3–5 persons, approximately 50% had a child/children under the age of 18 in the family, and 57.5% had an average monthly household income of more than ¥ 6000 Yuan. The basic demographics of the respondents were consistent with the demographics of Suzhou. Table 2 displays the statistical samples of our survey among 209 respondents in Suzhou, China. The data revealed that if a hypothetical food safety incident caused by abuse of food additives erupted, 92.8% of the respondent would consider the results serious, and 87.2% of the respondents would be angry at the incidence; 56% of the respondents would respond with extreme behaviors, such as a petition. These results were not surprising, as in recent years, food safety accidents caused by the abuse of food additives have occurred repeatedly in China. Moreover, 57.1% consumers had no or very little confidence in the food market. The findings in Table 2 reveal that safety incidents due to abuse of food additives caused the Chinese public’s scares of the domestic food market. Reliability and validity test The value of the Kaiser–Meyer–Olkin (KMO) measure, an assessment of the appropriateness of using factor analysis on data,

was calculated to be 0.7432; the approximate chi-square from Bartlett’s Test of Sphericity was 1554.535, and P-value was less than 0.01. Thus the null hypothesis3 was rejected. The results demonstrated that common factors existed among the primitive variables and that factor analysis was appropriate. Factor analysis and determination of the number of factors The number of factors to be retained was determined using a parallel analysis (Horn, 1965). The eigenvalues of a random matrix were computed using a command statement in SPSS18.0 (Hayton, Allen, & Scarpello, 2004), and then the mean was computed. Such two-step computations were repeated 50 times to obtain a set of average eigenvalues of a random matrix. The scree plot of actual eigenvalues and the mean of randomly generated eigenvalues were then compared. The absolute maximum number of factors to be extracted was determined according to the position of the point of intersection. As shown in Fig. 2, the actual eigenvalues and the mean of randomly generated eigenvalues intersect at the fourth eigenvalue. It was thus decided that four factors, which explained 66.783% of the total variance should be extracted. Therefore, the four common factors chosen for further analysis were: the attitudes toward the behavior, subjective norm, information perception, and risk perception. Oblique rotation was applied to determine the number of items included in each factor (Fabrigar, Wegener, MacCallum, & Strahan, 1999) in Mplus6.0. The rotated factor loading matrix is shown in Table 3. Two of the items, ‘‘unwilling to buy food containing additives because of the experience of the family’’ and ‘‘losing confidence in the domestic food market’’, cross loaded. In terms of 2 Kaiser (1974): The KMO test compares the simple and partial correlation coefficient between the variables. If the partial correlation coefficient is much less than the simple correlation coefficient, the correlation is higher, and the KMO value is close to 1. Generally, a KMO value of above 0.9, 0.8–0.9, 0.6–0.8, 0.5–0.6 and below 0.5 represents ‘‘very suitable’’, ‘‘suitable’’, ‘‘moderate’’, ‘‘not suitable’’ and ‘‘not suitable at all’’, respectively. 3 Petroni and Braglia (2000): Bartlett’s Test of Sphericity is based on the correlation matrix. Its null hypothesis is that the correlation matrix is an identity matrix.

94

L. Wu et al. / Appetite 70 (2013) 90–98 Table 4 Reliability and construct validity test. Item

WHOLE ATTI SN INMF RISK INTEN

Item number

Cronbach-

a

Guttman split-half

Number of common factors

Variance contribution rate (%)

18 3 4 4 4 3

0.819 0.728 0.741 0.801 0.776 0.734

0.710 0.704 0.780 0.726 0.734 0.656

– 1 1 1 1 1

– 58.923 56.631 63.663 61.186 57.160

Note: Please refer to Table 1 for abbreviations.

Fig. 2. Plot of actual versus randomly generated eigenvalues.

Table 3 Factor loadings for exploratory factor analysis with oblique rotation. Composition

Factor 1

Factor 2

Factor 3

Factor 4

FACI SYNI EXCE MEDI FRID REPU FAMI GOVE INEN INMD INPR BZXI INSY LCFI MORL HARM RGHC ISLF

0.712 0.817 0.385 0.067 0.010 0.002 0.078 0.181 0.125 0.124 0.001 0.008 0.016 0.088 0.080 0.031 0.097 0.009

0.043 0.051 0.106 0.436 0.608 0.535 0.314 0.506 0.142 0.024 0.108 0.115 0.039 0.115 0.040 0.008 0.360 0.156

0.042 0.014 0.050 0.102 0.125 0.064 0.075 0.039 0.434 0.675 0.595 0.867 0.749 0.125 0.023 0.274 0.037 0.098

0.013 0.035 0.017 0.101 0.142 0.197 0.339 0.126 0.069 0.023 0.019 0.021 0.026 0.413 0.839 0.541 0.314 0.508

Note: Factor loadings >0.30 are in boldface. Please refer to Table 1 for abbreviations.

actual meaning, these two items coincided with ‘‘influence from the family members or friends on buying food with additive’’ and ‘‘doubt about the safety of domestic food market’’ and were thus removed. In all, 16 items remained in further analyses.

Variable index determination The four common factors obtained in the factor analysis were subjected to reliability and validity tests. The reliability coefficients obtained from SPSS 18.0 were all within the reference value. Furthermore, a factor analysis was performed to evaluate the construct validity of the sample data, and the results are shown in Table 4. Table 4 reveals that there was only one common factor, and that the variance contribution rate and factor loading of the first common factor were greater than 0.5.4 This indicated that the four dimensions had good construct validity, which confirmed the valid construct of the dimensions of the hypothetical model, and verified the corresponding indicator variables. As shown in Fig. 3, the path diagram and path coefficients were obtained by using AMOS18.0 based on the four dimensions extracted by factor analysis, i.e., attitudes toward the behavior, 4 Nunnally (2010), Kerlinger and Lee (1973) and Hair, Black, Babin, Anderson, and Tatham (2010): Construct validity is the consistency between the common factor structure obtained by factor analysis and the questionnaire structure. When there is only one common factor, and the contribution rate and factor loading are greater than 0.5, the construct validity is considered good.

subjective norm, information perception, and risk perception as well as their respective identified items. Attitudes toward the behavior refer to the public’s positive or negative evaluation of food additive safety. It is composed of three items, ‘‘buying food with additives makes me uncomfortable due to the possibility of synthetic additives harming our health’’, ‘‘buying food with additives makes me worried because excessive addition of additives is harmful’’ and ‘‘buying food with additives makes me feel unpleasant even though it facilitates our lives’’. Subjective norm reflects the influence on the public risk perception regarding food additive safety and fears caused by family members, friends, media, and social organizations. It consists of four items, ‘‘ influence of media reports with regard to buying food with additives on the public’s behavior’’, ‘‘influence on buying food with additives from family members or friends’’, ‘‘influence on buying food with additives from government policies’’, and ‘‘influence on buying food with additives from enterprise reputation’’. Information perception describes the influence from public perception of information regarding food additives on consumer purchase intention. It is composed of four items, ‘‘influence from information regarding the presence or absence of quality and safety certification of food with additives’’, ‘‘influence from information of media reports regarding food with additives’’, ‘‘influence from information regarding experts’ opinions on food with additives’’, and ‘‘influence from information regarding additives on the package labels’’. Risk perception indicates the impact of public risk perception on their purchase intention; it is composed of four items, ‘‘doubt about the food safety of the domestic food market’’, ‘‘absence of a code of ethics in enterprises makes a person suspect the safety of the food with additives’’, ‘‘abuse of food additives became a major potential food risk’’, and ‘‘ineffective government regulation on additives’’. Variables included in the above four dimensions were all measured using a seven-point Likert scale. Higher score indicates a higher degree of agreement. As shown in Table 4, Cronbach’s alpha5 of attitudes toward the behavior, subjective norm, information perception, and risk perception was 0.728, 0.741, 0.801, and 0.776, respectively, which indicated that the internal consistency between the items was relatively high. Public trust in food with additives is generally lower after outbreaks of food safety incidents caused by additive abuse. The majority are unwilling to buy food with additives. Thus, three items, ‘‘would rather buy expensive food without additives than buy less expensive food with additives’’, ‘‘will buy food with additives only if the content of the additives strictly follows the national standards’’, and ‘‘will not buy food with additives under any circumstances,’’ were used to reflect the public’s intention to purchase food with additives under food scares. They were also measured using a seven-point Likert scale. The higher the score, 5 Guilford and Fruchter (1965): A Cronbach’s alpha greater than 0.7 indicates high reliability, while one less than 0.35 indicates low reliability and should be deleted. The Guttman split-half coefficient should usually be greater than 0.5.

95

L. Wu et al. / Appetite 70 (2013) 90–98

Fig. 3. Path coefficient diagram of SEM. P < 0.05,



P < 0.01,



P < 0.001.

Table 5 Demographic analysis on buying intention. Will buy food with additives unless it costs less money Agree

* **

Disagree

Total

Gender Male Female

v2 = 4.027* 71 (61%) 44 (47%)

45 (39%) 49 (53%)

116 93

Age <30 30–60 >60

v2 = 11.054*** 64 (75%) 55 (65%) 18 (45%)

21 (25%) 29 (35%) 22 (55%)

85 84 40

Income <6000 P6000

v2 = 4.323* 24 (27%) 49 (41%)

89 120

Education High school and below College and above

v2 = 1.032 25 (38%) 44 (31%)

66 143

65 (73%) 71 (59%) 41 (62%) 99 (69%)

Will buy food with additives only if content of additives follows standards Agree

Disagree

Total

v2 = 0.581

Agree

Disagree

Total

79 (68%) 66 (71%)

116 93

v2 = 0.001

74 (64%) 64 (69%)

42 (36%) 29 (31%)

116 93

31 (36%) 37 (44%) 14 (35%)

85 84 40

34 (38%) 39 (33%)

89 120

20 (30%) 69 (48%)

66 143

v2 = 1.389

37 (32%) 27 (29%)

v2 = 6.074**

54 (64%) 47 (56%) 26 (65%)

v2 = 0.731

49 (58%) 34 (40%) 16 (40%)

36 (42%) 50 (60%) 24 (60%)

85 84 40

55 (62%) 76 (64%)

89 120

42 (64%) 98 (69%)

66 143

v2 = 0.001

55 (62%) 81 (68%)

v2 = 5.950* 46 (70%) 74 (52%)

Will buy food with additives

34 (38%) 43 (36%)

v2 = 0.489 24 (36%) 45 (31%)

P < 0.05. P < 0.01. P < 0.001.

***

the lower the public’s intention to purchase food with additives, and the more serious the resulting scares of food additives would be. Cronbach’s alpha was 0.734, which indicated that the internal consistency between the variables was relatively high. Results Demographic analysis on buying intention As shown in Table 5, participants’ buying intention was affected by gender, income and education at various levels depending on the construct of the questions. Men were more sensitive to price

and more likely to buy food with additives for cost reduction reasons compared to women. Younger participants were more open to food with additives in general, while older participants were more cautious about additives and were less likely to buy food with additives for cost reduction reasons. Those participants with income less than ¥ 6000 Yuan also tended to accept food with additives for lower cost. Participants with higher education had more doubt about additives and less likely to buy food with additives even if the content of additives follows regulatory standards, while those with lower education tended to have more faith in the national or international standards.

96

L. Wu et al. / Appetite 70 (2013) 90–98

Parameter test and fitness evaluation

Table 7 Regression weight of structural model.

The fitting results demonstrated no offending estimates in the hypothetical model. Therefore, a test of overall model goodness of fit was performed. The results of the overall goodness of fit test for the hypothetical model were acceptable (Table 6). This demonstrated that the goodness of fit of the overall model was high. That is the causal model conformed to the actual survey data, and that the modified hypothetical model for path analysis was valid. Path analysis of the structural model As shown in Table 7, the standardized path coefficients from subjective norm to risk perception, the attitudes toward the behavior to risk perception, and the information perception to risk perception were 0.490, 0.431, and 0.732, respectively. The standardized path coefficients from subjective norm to purchase intention, the attitudes toward the behavior to purchase intention, and the information perception to purchase intention were 0.457, 0.464, and 0.521, respectively. The standardized path coefficient from risk perception to purchase intention was 0.757. The influence from subjective norm and information perception on risk perception, and that from information perception and risk perception on purchase intention both met the test of significance. These results showed that subjective norm and information perception had a significant positive impact on risk perception, and that information perception and risk perception also significantly influenced purchase intention. A. The standardized path coefficient from risk perception to purchase intention was the highest, indicating that the greater the risk the public perceives, the more serious the resulting scare, and the lower their intention to purchase. This result is consistent with the findings of Lobb (2005) and supported hypothesis H7. B. The standardized path coefficients from subjective norm to risk perception and purchase intention were 0.490 and 0.457, respectively, of which, the former met the test of significance at 95% confidence interval, and the latter one was significant at 90% confidence interval, supporting hypothesis H3 and H4. This result demonstrated that, with little knowledge of food additives, public risk perception of food additives was influenced by media, family members, friends, and the reputation of enterprises. This finding is consistent with the findings of Verbeke and Kenhove (2002). C. The standardized path coefficients from attitudes toward the behavior to risk perception and purchase intention were 0.431 and 0.464, respectively. The effect is close to significance (P = 0.052, 0.058), indicating that attitudes toward the behavior had some influence on risk perception and purchase intention. This finding is consistent with the findings of Chen and Li (2007).

Table 6 The evaluating standard and the evaluating result of the integral fitness of SEM. Index name

Evaluating standard

Actual fitting value

Result

GFI RMSEA AGFI NFI IFI TLI CFI PNFI PCFI PGFI v2 DOF ratio

>0.90 <0.06 >0.90 >0.90 >0.90 >0.90 >0.90 >0.50 >0.50 >0.50 <2.00

0.913 0.051 0.874 0.837 0.936 0.914 0.934 0.646 0.720 0.630 1.539

Ideal Ideal Close Close Ideal Ideal Ideal Ideal Ideal Ideal Ideal

Path

Parameter estimated value

Standard error

Standardized regression weights

P value

RISK SN RISK ATTI RISK INFM INTEN SN INTEN ATTI INREN INFM INTEN RISK

0.528 0.536 0.821 0.292 0.273 0.438 0.358

0.181 0.276 0.372 0.239 0.267 0.222 0.135

0.490 0.431 0.732 0.457 0.464 0.521 0.757

0.004 0.052 0.000 0.077 0.058 0.023 0.000

Note: Please refer to Table 1 for abbreviations.

Table 8 Factor loading analysis of the measurement models. Path GOVE FRID REPU MEDI BZXI INPR INMD INEN EXCE FACI SYNT MORL HARM LCFI ISLF NEVE STAN MORE

SN SN SN SN INFM INFM INFM INFM ATTI ATTI ATTI RISK RISK RISK RISK INTEN INTEN INTEN

Parameter estimated value

Standard error

Standardized regression weights

P value

1.000 0.494 1.982 2.047 1.000 1.522 0.819 1.212 1.000 1.032 1.197 1.000 0.824 0.921 1.264 1.000 2.147 1.911

– 0.187 0.344 0.335 – 0.328 0.237 0.265 – 0.221 0.247 – 0.131 0.137 0.170 – 0.613 0.545

0.437 0.217 0.825 0.811 0.395 0.602 0.338 0.581 0.503 0.498 0.534 0.596 0.562 0.615 0.721 0.277 0.810 0.728

– 0.008 0.000 0.000 – 0.000 0.000 0.000 – 0.000 0.000 – 0.000 – 0.000 – 0.003 –

Note: The four paths with ‘‘–’’ indicate the benchmark for parameter estimation in SEM. Please refer to Table 1 for abreviations.

D. The standardized path coefficients from information perception to risk perception and purchase intention were 0.732 and 0.521, respectively, supporting hypothesis H5 and H6. Public perception of information on food additives had a significant impact on their risk perception and purchase intention, which is consistent with the conclusions of Shim et al. (2011). Factor loading analysis of the measurement models (Table 8) The factor loadings reflect the level of influence of the measurable variable on the latent variables. The model fitting results revealed that: A. The standardized coefficients of food enterprise reputation and media reports about the public’s behavior were 0.825 and 0.811, respectively. They were the two most prominent features in the latent variable subjective norm, a finding that was consistent with the findings of Zhang (2009). Because of the universality and extensive coverage of media reports, including online media reports, on food safety incidents in China, they are very likely to affect the public’s awareness of food safety incidents. Jointly with the influence of the enterprise reputation, media coverage may trigger public food scares. B. The standardized coefficients of information on experts’ opinions regarding food with additives and information regarding the presence or absence of food quality and safety certification were 0.602 and 0.581, respectively. They were

L. Wu et al. / Appetite 70 (2013) 90–98 Table 9 Covariance among exogenous variables. Path

Parameter Estimated Value

S.E.

C.R.

P value

INFM M AATI SN M INFM SN M ATTI

0.341 0.207 0.104

0.099 0.068 0.085

3.447 3.046 2.374

0.000 0.000 0.020

Note: Please refer to Table 1 for abbreviations.

the two most prominent features in the latent variable information perception. These results demonstrated that the information on experts’ opinions regarding food with additives and information regarding the presence or absence of food quality and safety certification significantly affected the public risk perception and jointly acted on their purchase intention. These findings are consistent with the findings of Gailliot et al. (2007) and Mazzocchi et al. (2008). C. The standardized coefficient of evaluation of the government’s regulatory strength on additive usage was 0.721. It was the most prominent feature in the latent variable of risk perception. This indicated that when food additive safety incidents erupted, the public’s perceived risk was significantly influenced by the government’s regulatory effectiveness. Therefore, the weaker the government regulation of food safety is considered to be, the greater the food safety risk would be perceived, and thus the higher likelihood of suffering from food scares. These findings are consistent with the findings of Rojas and Brewer (2007). Covariance between the exogenous latent variables The estimates of covariance between the exogenous latent variables are summarized in Table 9. Covariance between attitudes toward the behavior, subjective norm, and information perception were significant. Thus, public perception of family and friends’ preferences of food additives, public perception of information regarding food additives, their attitudes toward food additives and the public information perception of food additives co-vary. Thus, hypothesis H8 was supported. Discussion Main conclusions and policy implications In this study, the proposed model was used to explore public risk perception of food additive safety and food scares. The results revealed that subjective norm and information perception were the main factors influencing the public risk perception of food additives and the resulting food scares. Information perception exerted the most significant effect on the public’s purchase intention. In addition, the subjective norm and information perception also had a significant effect. Risk perception was affected by information perception and subjective norm. Therefore, information perception, subjective norm, and information perception all affected the public’s purchase intention. However, attitudes toward the behavior had a marginal (P = 0.052) effect on public risk perception of food additives and food scares. Although the influence between attitude toward behavior and public risk is not obvious, the effect is close to significance. This may be a result of the low statistical power and the small sample size of this study. The results of demographic analysis showed that lower cost, a potential consumer benefit, has positive impact on men, younger generation, and low income individuals. Consider most of the purchasing decisions are made by women, the effect of reducing cost

97

by adding food additives may not result in much change in consumer behaviors in the society. Interestingly, people with lower levels of education were more likely to purchase food with additives that follows government standards than those with higher levels of education. This suggests that consumers with lower levels of education are more likely to trust government institutions to regulate food additives. Hence, to reduce general public’s food scares, strengthen government regulation or communicating through government authorities may have positive impact on people with low education, which is the majority of the Chinese population. Chinese society is now experiencing a significant transformation, with various social contradictions blending together. The conclusions of this study provide some reference to the governments of cities with a relatively high economic level to prevent public food scares in China. Whether some safety issues develop into fully-fledged food scares or not, depends on the magnitude of risk faced by public, as well as the extent of media attention devoted to that specific food safety issue (Knowles et al., 2007). At present, the dissemination of online public opinion on food safety has resulted in very serious problems and has had a negative impact in China. Public scares are an objective reflection of a major crisis in the society, but they have adverse impacts on the entire social crisis management if not controlled in a timely fashion (Li et al., 2011; Yan, 2011). Therefore, the joint efforts of the government and the public are required to improve the public risk perception of food additives, which has a fundamental role in resolving possible food scares, as well as preventing and controlling them. Studies have shown that participants’ suspicion of food additives approved by the government was derived from insufficient information and misunderstandings of food additives, as well as a lack of clarity in risk communications among the stakeholders such as the government, industry, and consumers (Shim et al., 2011). It is therefore imperative to establish an effective mechanism of food safety risk communication, to disseminate information of food safety risk scientifically, to curb the media’s misleading reports, to pay close attention to the publicity and randomness of online public opinion on food safety and the possible resulting crisis, and to avoid triggering harmful chain reactions that are difficult to control. As people with lower levels of education are more likely to trust government instructions, it is critical to release information of government supervision on food safety and related efforts in a timely manner. Finally, it is important to promote popularization and education of basic knowledge of food safety and enhance the public knowledge about food additives in China. All of these are important to restore public confidence in the food market and gradually increase the public’s expected controllability of scares.

Prospects This study demonstrated that subjective norm and information perception were the major factors affecting the public risk perception of food additives, which in turn had a significant impact on the public’s purchase intention. This finding suggested that, after the outbreak of food safety incidents caused by abuse of additives in China, the public risk perception of food additives was easily influenced by behavior of reference groups and a variety of information sources. However, this study only confirmed the influence of the public information perception on their risk perception and food scares. It did not explore the influence of the public trust on the information regarding food scares nor did it compare the influence of information from different sources. In order to deepen the study of public food scares caused by abuse of food additives, comparison of the influence of information framing and information

98

L. Wu et al. / Appetite 70 (2013) 90–98

sources on public food scares can be considered for inclusion in a follow-up study. In addition, due to the limitation on sample size, this study did not modify the model by including demographic variables. However, the results revealed the significance of those variables, and it is suggested that future studies on this topic in China should systematically explore the effects of demographics. The outcome could help design communications tailored to different groups to mitigate negative psychological and behavioral responses on food containing additives.

References Ajzen, I. (2000). Construction of a standard questionnaire for the theory of planned behavior. Retrieved 20.11.00. Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predicting social behavior. Englewood Cliffs, NJ: Prentice-Hall. Chen, M. F., & Li, H. L. (2007). The consumer’s attitude toward genetically modified foods in Taiwan. Food Quality and Preference, 18(4), 662–674. Cook, A. J., Kerr, G. N., & Moore, K. (2002). Attitudes and intentions towards purchasing GM food. Journal of Economic Psychology, 23(5), 557–572. E, P. (2012). The top ten issues attracting Chinese public attention in 2012. Food safety is in the first place of the list. Insight China, 12, 21–26. Fabrigar, L. R., Wegener, D. T., MacCallum, R. C., & Strahan, E. J. (1999). Evaluating the use of exploratory factor analysis in psychological research. Psychological Methods, 4, 272–299. Falguera, V., Aliguer, N., & Falguera, M. (2012). An integrated approach to current trends in food consumption. Moving toward functional and organic products. Food Control, 26(2), 274–281. Fischer, A. R., De Jong, A. E., De Jonge, R., Frewer, L. J., & Nauta, M. J. (2005). Improving food safety in the domestic environment. The need for a transdisciplinary approach. Risk Analysis, 25(3), 503–517. Fishbein, M. (1963). An investigation of the relationships between beliefs about an object and the attitude toward that object. Human Relations, 16(3), 233–240. Fu, G. Q., & Tong, X. Y. (2003). Brand, price and product origin’s influence in consumers’ purchase choice. Journal of Management Science, 12(6), 80–84. Gailliot, M. T., Baumeister, R. F., DeWall, C. N., Maner, J. K., Plant, E. A., Tice, D. M., et al. (2007). Self-control relies on glucose as a limited energy source. Willpower is more than a metaphor. Journal of Personality and Social Psychology, 92(2), 325–336. Graham, J. W. (2009). Missing data analysis. Making it work in the real world. Annual Review of Psychology, 60, 549–576. Graham, J. W., Olchowski, A. E., & Gilreath, T. D. (2007). How many imputations are really needed? Some practical clarifications of multiple imputation theory. Prevention Science, 8(3), 206–213. Guilford, J. P., & Fruchter, B. (1965). Fundamental statistics in psychology and education. New York: McGraw-Hill. Hair, J. F., Black, W. C., Babin, B. J., Anderson, R. E., & Tatham, R. L. (2010). Multivariate data analysis. Upper Saddle River, NJ: Prentice Hall. Hayton, J. C., Allen, D. G., & Scarpello, V. (2004). Factor retention decisions in exploratory factor analysis. A tutorial on parallel analysis. Organizational Research Methods, 7(2), 191–205. Horn, J. L. (1965). A rationale and test for the number of factors in factor analysis. Psychometrika, 32, 179–185. Kaiser, H. F. (1974). An index of factorial simplicity. Psychometrika, 39(1), 31–36. Kerlinger, F. N., & Lee, H. B. (1973). Foundations of behavioral research. New York: Holt, Rhinehart and Winston. Knowles, T., Moody, R., & McEachern, M. G. (2007). European food scares and their impact on EU food policy. British Food Journal, 109(1), 43–67. Kothe, E. J., Mullan, B. A., & Butow, P. (2012). Promoting fruit and vegetable consumption. Testing an intervention based on the theory of planned behavior. Appetite, 58, 997–1004. Li, X. (2011). Cognitive and regulatory food safety risks in the risk society perspective. Administrative and Law, 6, 25–28. Li, Q., Liu, W., Wang, J., & Dai, Y. (2011). Application of content analysis in food safety reports on the Internet in China. Food Control, 22(2), 252–256. Liu, Y., & Zhang, F. (2004). China’s higher education of empirical research to pay and tuition policy. China Soft Science, 2, 14–20. Lobb, A. (2005). Consumer trust, risk and food safety. A review. Food Economics-Acta Agriculturae Scandinavica, Section C, 2(1), 3–12.

Lobb, A. E., Mazzocchi, M., Traill, W. B. (2006). Risk perception and chicken consumption in the avian flu age. A consumer behaviour study on food safety information. In Annual meeting of American agricultural economics association (July). Lobb, A. E., Mazzocchi, M., & Traill, W. B. (2007). Modelling risk perception and trust in food safety information within the theory of planned behaviour. Food Quality and Preference, 18(2), 384–395. Mazzocchi, M., Lobb, A., Bruce Traill, W., & Cavicchi, A. (2008). Food scares and trust. A European study. Journal of Agricultural Economics, 59(1), 2–24. McGuire, W. J. (1969). The nature of attitudes and attitude change. The Handbook of Social Psychology, 3, 233–346. Meiselman, H. L., & MacFie, H. J. (1996). Food choice acceptance and consumption. Blackie Academic & Professional. Moschitz, H., & Stolze, M. (2010). The influence of policy networks on policy output. A comparison of organic farming policy in the Czech Republic and Poland. Food Policy, 35(3), 247–255. Nunnally, J. C. (2010). Psychometric theory 3E. Tata McGraw-Hill Education. Ouyang, H. Y. (2011). Nearly 70 percent of the respondents felt insecure about food. 2010–2011 Report on consumer confidence in food safety. Insight China, 1, 42–45. Petroni, A., & Braglia, M. (2000). Vendor selection using principal component analysis. Journal of Supply Chain Management, 36(2), 63–69. Prati, G., Pietrantoni, L., & Zani, B. (2012). The prediction of intention to consume genetically modified food. Test of an integrated psychosocial model. Food Quality and Preference, 25(2), 163–170. Qin, W., & Brown, J. L. (2008). Factors explaining male/female differences in attitudes and purchase intention toward genetically engineered salmon. Journal of Consumer Behavior, 7, 127–145. Rojas, M. C., & Brewer, M. S. (2007). Effect of natural antioxidants on oxidative stability of cooked, refrigerated beef and pork. Journal of Food Science, 72(4), 282–288. Schafer, J. L., & Graham, J. W. (2002). Missing data. Our view of the state of the art. Psychological Methods, 7(2), 147–177. Sharlin, H. I. (1987). Macro-risks, micro-risks, and the media. The EDB case. In The social and cultural construction of risk (pp. 183–197). Netherlands: Springer. Shim, S. M., Seo, S. H., Lee, Y., Moon, G. I., Kim, M. S., & Park, J. H. (2011). Consumers’ knowledge and safety perceptions of food additives. Evaluation on the effectiveness of transmitting information on preservatives. Food Control, 22(7), 1054–1060. Slovic, P. (1993). Perceived risk, trust and democracy. Risk Analysis, 13(6), 675–682. Stern, T., Haas, R., & Meixner, O. (2009). Consumer acceptance of wood-based food additives. British Food Journal, 111(2), 179–195. Verbeke, W., Frewer, L. J., Scholderer, J., & De Brabander, H. F. (2007). Why consumers behave as they do with respect to food safety and risk information. Analytica Chimica Acta, 586(1), 2–7. Verbeke, W., & Kenhove, P. V. (2002). Impact of emotional stability and attitude on consumption decisions under risk. The Coca-cola crisis in Belgium. Journal of Health Communication, 7(5), 455–472. Verbeke, W., & Ward, R. W. (2001). A fresh meat almost ideal demand system incorporating negative TV press and advertising impact. Agricultural Economics, 25(2–3), 359–374. Wang, S. T. (2010). An examination of the relationship among extrinsic cue, perceived value, trust and word-of-mouth. Using online popular food as an example (pp. 91–93). National Pingtung Institute of Commerce. Williams, P. G., Stirling, E., Keynes, N. (2004). Food fears. A national survey on the attitudes of Australian adults about the safety and quality of food (pp. 32–39). Faculty of Health and Behavioral Sciences. Papers, 13. Worsley, A. K. (1996). Which nutrition information do shoppers want on food labels? Asia Pacific Journal of Clinical Nutrition, 5, 70–78. Worsley, A., Worsley, A., & McConnon, S. (1991). Kiwis, food and cholesterol. New Zealand consumers’ food concerns and awareness of nutritional guidelines. Australian Journal of Public Health, 15(4), 296–300. Wu, L. H., & Huang, W. D. (2012). Introduction to 2012 China development report on online public opinion of food safety. Beijing: People’s Press. Wu, L. H., Zhang, Q. Q., Shan, L. J., & Chen, Z. X. (2012). Study on identifying critical factors of using behavior of food additives in enterprises. A fuzzy DEMATEL method. Systems Engineering, 30(7), 48–54. Yan, J. (2011). On network moral ethics construction from the perspective of public opinion about food safety. Journal of Henan University of Technology, 7(2), 22–25. Zagata, L. (2012). Consumers’ beliefs and behavioral intentions towards organic food. Evidence from the Czech Republic. Appetite, 59(1), 81–89. Zhang, L. L. (2009). Negative effects of food safety report and supervision by public and psychological appease, the case of ‘‘Sanlu milk powder incident’’ and ‘‘maggots oranges event’’. Journal of Luoyang Normal University, 28(2), 101–104.