Consumer Response to Food Safety Events: An Interaction Between Risk Perception and
Trust of Information in the Chicken and Beef Markets
Jonathan D. Shepherd and Sayed Saghaian
University of Kentucky
Department of Agricultural Economics
314 Charles E. Barnhart Building
Lexington, KY 40546-0276
Phone: (859) 257-9710 ext. 272
E-mail: [email protected]
Phone: (859) 257-2356
Selected Paper prepared for presentation at the Southern Agricultural Economics
Association Annual Meeting, Dallas, TX, February 2-6, 2008
Copyright 2008 by Jonathan D. Shepherd and Sayed Saghaian. All rights reserved.
Readers may make verbatim copies of this document for non-commercial purposes by any
means, provided that this copyright notice appears on all such copies.
Recent food safety events have captured substantial media attention, increased
consumers’ awareness of food safety concerns and further complicated marketing aspects
of agricultural products today. Economic losses associated with such events are not
limited to the immediate time period following an occurrence, but potentially have longrun effects and reach beyond local and domestic markets. Food safety events can open
competitive opportunities for individual firms within an affected industry to differentiate
their products’ attributes, marketing safer production methods in an attempt to capture a
larger share of the market (Bruhn and Schutz, 1999). Another challenging consequence
of food safety events is the potential loss associated with international markets.
Oftentimes, countries will ban or limit imports of certain products from countries facing a
food safety occurrence. For example, Japan banned US imports of beef following a
Bovine Spongiform Encephalopathy (BSE) outbreak, creating a barrier for the US beef
industry to overcome in a country with exceptional quality differentiation standards
(Saghaian and Reed, 2004). Further, research indicates that consumers consider all food
safety concerns (i.e. genetic modification, hormone/antibiotic use, E. coli/salmonella) in
their decision to purchase agricultural products, highlighting the importance for decision
makers to understand how society perceives risks associated with foods (Bruhn and
In the United States this year, there have been E. coli outbreaks in the ground beef
industry and a Clostridium botulinum outbreak that occurred in processed canned meats
that alone accounted for over 721,000 pounds (USDA, 2007). Also, there has been
concern over Avian Influenza after reported outbreaks in three US states and in Canada in
2004 (CDC, 2006). Food safety events and their impacts have been extensively
investigated in the literature. The results of these studies generally show that food safety
events affect demand adversely (Henson and Northern, 2000). Many studies also focus
on willingness to pay for reduced chances of food safety events, and it also has been
shown how society trusts information from governing bodies with regards to said events
(Smith, Ravenswaaye and Thompson 1998; Henson, 1996).
The life cycle of a food safety event is a dynamic process where consumers often
change consumption patterns during the scare, returning to pre-scare consumption
patterns after the event. It is unclear how long the cycle takes or what signals are most
effective to persuade consumers back their pre-scare consumption habits. Sociological
researchers argue that, generally, a food safety event receives prominent media coverage
with consumers initially over-reacting by avoiding the identified food item (Mazzocchi,
Stefani, and Henson, 2004). Media coverage of food safety events can also be confusing
to consumers as more and more of the information is revealed to the public because of
time lapses in coverage or conflicting information within or between different media
sources (Caswell, 2006). This is of particular concern to affected firms, as consumers
often rely primarily on media coverage for information concerning such events (Wade
and Conley, 1999).
Economic impacts of food safety events vary greatly from incident to incident.
Topps Meat Company suffered the second largest meat recall in US history for E. coli
contaminated ground beef in 2007. An October 6, 2007 New York Times article reported
that Topps Meat Company had to shut down operations as a result of the recall. The
article also mentioned the chief operating officer for the company lamenting that the
scale of the recall was too large to recover the business losses. Although businesses
closing as a result of a food safety event may not be common, substantial effort is
required on the part of the firm to restore consumer confidence. The Mexican fast-food
chain, Taco Bell, faced a daunting marketing recovery task in 2006, after an E. coli
outbreak linked to lettuce (Taco Bell). The company reacted quickly with television
commercials and governmental voices to reassure consumers that the situation was being
handled and that it was safe to eat again.
Noting economic theory, food safety events will negatively affect demand for
products involved in the immediate time period following the crisis. However, long term
effects are not as clear as consumers may turn to other perceived safer products
(McCluskey et al, 2005). This may be realized by consumers substituting to other brands
within the same industry or substituting completely to a different product all together.
Food safety events are more complicated than other risky endeavors as an absolute
reduction in risk is not possible because food is essential to life, eradicating the
possibility of a complete reduction in food safety risk (Frewer et al, 1998). Another
complicating issue is that food choice is a personal decision, often solidified by a
person’s past, and results in quickly realized benefits of food consumption (Fife-Schaw
and Rowe, 1996). This means there is a potential for food scares to have economic
impacts from food purchasing decisions of generations to come without effective
In the mid 1990’s the EU (European Union) experienced a BSE outbreak that
resulted in a decline in the demand for beef as a whole. However, some individuals
actually increased their demands (Henson and Northern, 2000). Exceptions like these
shed light on the dynamics of food safety events, and imposes the need for governments
and producers to understand how society conceptualizes food risk in order to have
effective policies (Lobb, Mazzocchi, and Traill, 2006).
The data for this research was obtained from a 2,000 random household sample
that targeted heads of households in the five counties that contained the five largest cities
in Kentucky. The survey was conducted via United States Postal Service and a $2 “token
of appreciation” was offered to respondents upon receiving a completed survey in an
attempt to ensure an adequate response rate. The survey instrument used was originally
developed by Lobb, Mazzocchi, and Traill (2006) with changes made to better fit our
area of interest and target population. The survey instrument contained 63 questions
most of which were measured on a 7-point Likert scale.
The objective of this paper is to examine the impact of food safety events in the
chicken and beef markets in Kentucky. This is achieved using the SPARTA model
(Lobb, Mazzochi, and Traill, 2007) based on the Theory of Planned Behavior (TPB)
developed by Ajzen (1991). Of particular interest is determining whom consumers trust
with regards to information concerning a food safety event, and what other factors (social
or demographic) affect consumers’ response to such events. These results are compared
to the results of a EU study that focused on the perceived risks associated with chicken
consumption to see if the results can be generalized across different countries, regions
The SPARTA model represents subjective norms, perceived behavioral control,
risk, trust, and “alia” (all other variables) (Lobb, Mazzochi, Traill, 2007). See pictorial
representation, Figure 1. TPB is an extension of the Theory of Reasoned Action and links
attitude and beliefs to actions through intentions (Ajzen 1991). This approach has been
used in several studies, including the meat market in the UK (McEachern and Shroder,
2004), as well as evaluating food choices of adolescents (Dennison and Shepherd, 1995).
The first three variables S, P, and A are formulated under Fishbein and Ajzens’ (1976)
expectancy value formulation.
Following Lobb, Mazzocchi, and Traill (2007) the
construction of the variables appear below:
S ∝ ∑n jmj
Where ni and mi are normative beliefs and motivations to comply, respectively.
P ∝ ∑ ck p k
Where ci are control beliefs and pi are power of control beliefs.
A ∝ ∑ bi ei
Where bi behavioral beliefs and ei are outcome evaluations of these beliefs.
The risk component, R, is formed similarly to the variables above using the expectancyvalue formulation (Lobb, Mazzocchi, and Traill, 2007):
R ∝ ∑ rl k l
where ri are specific risk factors and ki are weights given by respondents stating their
given knowledge of each risk factor.
Principal component analysis with varimax rotation was used to account for
correlations that may exist between 19 entities respondents rated with regards to their
trustworthiness to form the T component of the model (Lobb, Mazzocchi, and Traill,
This reduced the number of variables in this component into 4 categories,
Suppliers, Gov’t/University, Organizations, and Media; T1, T2, T3, and T4 respectively.
The Suppliers category includes shopkeepers, supermarkets, organic shops, and
processors. All of these categories seem to cover the same concept of where a consumer
may obtain a food product.
The Gov’t/University category contains doctor/health
authority, university scientist, USDA, state and federal government.
categories are all entities that consumers would most likely consider possessing an
authoritative/policy influencing voice.
Organizations contain the sub-categories of
political groups, environmental and animal welfare organizations as well as television
On first glance, television documentary sub-category seems non
applicable. However, there is a common thread among the sub-categories in that they all
have a primary focus or cause.
Arguably, television documentaries focus on one
subject/cause, allowing its inclusion into this category.
Lastly, the Media category
contains typical forms of communication, newspaper, internet, radio, magazines, and
product label (Table 1).
Following Lobb, Mazzocchi, and Trail (2007) the trust component is as follows:
Tz = ∑α zs t s , z = 1,..., Z
where ts are the specific trust factors, α zs are the loading factors and T is the principal
component score where Z is the total number of components measured across.
Following Lobb, Mazzocchi, and Traill (2007), four models were estimated;
consumers’ intention to purchase chicken and/or beef next week in general (ITP11) and
consumers’ intention to purchase chicken and/or beef next week given a hypothetical E.
coli/salmonella outbreak (ITP21).
Theses models were also estimated using socio-
demographic variables to determine if such variances have an effect on the probability of
purchasing decisions (ITP12 and ITP22 respectively). An ordered probit regression was
used to estimate these models because of the ordered structure of the data and appears
below (Lobb, Mazzocchi and Traill, 2007):
I b = β 0 + β 1 S + β 2 P + β 3 A + β 4 R + ∑ λ z Tz
The inclusion of socio-demographic variables used for models ITP12 and ITP22 is as
I b = β 0 + ∑ γ 0i Di + β1 + ∑ γ 1i Di S + β 2 + ∑ γ 2i Di P + β 3 + ∑ γ 3i Di A +
β 4 + ∑ γ 4i Di R + ∑ λZ + ∑ γ gi Di TZ
Where SD is the ith socio-demographic variable
224 completed surveys were received, resulting in an 11.2% response rate.
Female response rate was 58% which is close to the 60% female response rate found by
Lobb, Mazzocchi and Traill (2006). The magnitude of this response rate is as expected
because females are still the principle food purchasers in many households (Lobb,
Mazzocchi, and Traill, 2006).
The number of people in the household had a minimum of 1 and a maximum of 7
with an average of 2.38. Average age of respondents was 54.45 year with a minimum of
20 and a maximum of 97 (Table 2).
69% of respondents reported some college
However, Lobb, Mazzocchi, and Traill (2006) found a slightly higher
percentage, roughly 72%. This magnitude is also as expected as I hypothesize that
individuals with higher education levels are better able to relate to the necessity of such
Respondents were asked to report their level of trust with information on a 7-point
Likert scale with regards to different entities that had hypothetically provided information
about potential risks associated with E. coil/salmonella in food. Political groups received
the highest percentage of completely distrust with over 17% of respondents choosing a
value of 1 on the scale. Governmental or political groups are often not be trusted by
consumers as these groups are seen by society as having a vested interest in protecting
firms (Frewer et al, 1996). The second highest value that was associated with completely
distrust was animal welfare organizations, with almost 14% of respondents choosing a
value of 1 on the scale. This also follows the idea that organizations such as these may
be perceived to have an agenda that biases the information they report.
authority, university scientists, and the USDA, were the top categories for which
consumers chose a level of 7, or complete trust, with 47.8%, 33% and 35.7% of
respondents choosing these categories respectively.
To elicit whom consumers trust as informational sources following a food safety
event, respondents were asked to assume they had heard rumors about a food safety
event. The survey instrument then had pairs of information sources and respondents were
asked whom they trusted more between each respective pair. 75.9% reported they trusted
university scientists over media and 74.1% reported they trusted university scientists over
producers. 70.5% reported trusting public authorities more than producers. This is of
interest to agribusiness firms. Establishing a representative from one of these groups
could help restore consumer confidence more quickly when communicating on a food
safety event (Table 3). Respondents were asked to indicate what media sources they
typically resort to in the face of a food safety event. Television accounted for over 33%
of responses, followed by internet at 27% and newspapers at 15%.
Respondents were asked to state their level of agreement with statements that
finished “My decision whether or not to buy chicken and/or beef next week is based on
the fact:.” Almost 52% of respondents chose a level 7 (complete agreement) or a level 6,
when prompted with the statement: “chicken and/or beef is a safe food.” With the same
statement, all responses greater than the level 4 (neither) account for almost 71% of
respondents (Table 4). When considering recent foods safety events occurring in these
markets, these results go against intuition. However, when considering demand for these
products has not suffered a steady decline, the results from here seem supported.
40.6% of respondents reported that it would be extremely unlikely that they
would purchase chicken and/or beef next week, if they had read an article in the
newspaper that high rates of E. coli/salmonella in chicken and/or beef had been found in
their area, resulting in several people being hospitalized. These results further solidify
the short-run impacts of food safety events extensive in the literature.
Respondents were prompted with a statement concerning their actions such as
proper food storage, handling, preparations, choice of place of purchase, and purchasing
higher quality products with regards to reducing the risks associated with food safety
events. About 51% of respondents stated their actions such as listed above would reduce
food safety risk by a large extent (value of 7) and 34% choose a value of 6, accounting
for 85% when summed. If all values over 4 (neither) were summed the total would be
93.8%. This has interest to food firms and the guidelines set forth by the Center for
Disease Control and their attempts to provide information to consumers about their part
in reducing food safety risks.
In the model concerning consumers’ intention to purchase chicken and/or beef
next week in general (ITP11), perceived behavioral control, trust in suppliers, and trust in
media had the largest negative impact on the probability of purchasing next week. Only
the perceived behavioral control variable was statistically significant.
Gov’t/University was positive and had the largest overall absolute impact on the
probability of purchasing. Subjective norms was the only other positive parameter. The
second model, consumers’ intention to purchase chicken and/or beef next week after a
hypothetical E. coli/salmonella event (ITP12), resulted in Trust in Gov’t/University
having the largest absolute impact, but in the opposite direction from the first model and
statistically significant. This can be interpreted as distrust in these sources positively
influence the probability to purchase or trust in these sources has a negative impact on he
probability. I expected this parameter to have a positive sign. It may be the case that
consumers do in fact associate this category as not being trustworthy contrary to
Media changed in absolute magnitude (by 0.05), became
positive and statistically significant. This means that trust in media increases the
probability of purchasing next week. Attitude remained unchanged in both magnitude
and direction. The other parameters only changed slightly in absolute magnitude but
most did change from negative to positive (Table 5).
The third model, consumers’ intention to purchase chicken and/or beef next week
following a hypothetical E. coli/salmonella food safety event (ITP21), did not result in
any statistically significant parameters. However, the sign and magnitude associated with
the subjective norm parameter (-0.02) was the same between this study and what was
found by Lobb, Mazzocchi, and Traill (2007). The other parameters consistent between
the two studies were attitude and risk perception and had the same direction but of
The final model, consumers’ intention to purchase chicken and/or beef next week
following a hypothetical E. coli/salmonella food safety event that also included sociodemographic variables (ITP22), resulted in 7 variables being statistically significant. Risk
perception was positive which was not as expected.
Risk coupled with the socio-
demographic shifter age and coupled with income had a negative sign associated with
them. These results are more plausible. It may be the case that increases in age and risk
perception would negatively affect the probability of purchasing a product following a
food safety event as the elderly are more likely to have worse complications in the face of
pathogenic contraction. Income was expected to have a negative effect on the probability
of a consumer purchasing a product from an industry that was facing a food safety event,
as relatively wealthier people are more able to completely substitute away from the
affected product to minimize risks. The Gov’t/University trust component with sociodemographic shifter education had a positive sign. This may be because relatively more
educated consumers may be better able to decipher through the media hype following
food safety events and turn towards more non-biased sources. Lobb, Mazzocchi, and
Traill (2007) found a positive association with this combination, but of larger magnitude,
0.08 compared to 0.02. Finally media with socio-demographic shifter education has a
negative sign associated with it. The intuitive explanation for this result is that increases
in education arguably make consumers more informed about potential bias in the media;
therefore, it would likely decrease the probability to purchase from a food safety event
affected market. Media and the shifter income are positive. It seems probable that
relatively higher income consumers would be better able to cross-reference media
coverage of a food safety event across multiple sources (Table 6).
Interesting conclusions can be drawn from the preliminary results of this research.
It is clear that consumers have established perceptions of the trust associated with
potential food safety information sources. Policy and decision makers alike can use these
results to better determine what media is used to convey food safety information. Also, it
is clear that consumers know that their actions can significantly reduce risks associated
with food safety. This is also of importance to firms that may want to increase the
education associated with these types of communication. Although interesting
conclusions can be drawn, I caution serious application of the empirical results from the
models other than the ITP22 because there of few significant factors in the other models.
Also, directional effects of the parameter estimates were not as expected in some models
as well. From this research it is not clear that consumers’ reactions can be generalized
across different regions or products. More research is needed in this area over more
subjects and products before generalizations can be made.
Ajzen, I. (1991). “The Theory of Planned Behavior.” Organizational Behavior and
Human Decision Processes. 50:179-211.
Belson, K. and Fahim, K. (2006). “After Extensive Beef Recall, Topps Goes Out of
Business.” The New York Times. 6 Oct. 2007.
Bruhn, C., and Schutz, H. (1999). “Consumer Food Safety Knowledge and Practices.”
Journal of Food Safety. 19: 73-87.
Caswell, J. (2006). “A Food Scare A Day: Why Aren’t We Better At Managing Dietary
Risk?” Human and Ecological Risk Assessment. 12: 9-17.
Center for Disease Control and Prevention. (2006). “Avian Influenza (Bird Flu): Past
Avian Influenza Outbreaks.” <http://www.cdc.gov/flu/avian/geninfo/pdf/avian_facts.pdf>
Dennison, C., and Shepherd, R. (1995) “Adolescent Food Choice: an application of the
Theory of Planned Behavior.” Journal of Human Nutrition and Dietetics. 8:9-23.
Fife-Schaw, C. and Rowe, G. (1996). “Public Perceptions of Everyday Food Hazards: A
Psychometric Study.” Risk Analysis. 16(4): 487-500.
Fishbein, M., and Ajzen, I. (1976). “Misconceptions about the Fishbein model:
Reflections on a study by Songer-Nocks. Journal of Experimental Social
Frewer, L., Howard, C., Hedderly,D., and Shepherd, R. (1998) “Methodological
Approaches to Assessing Risk Perceptions Associated With Food-Related
Hazards.” Risk Analysis. 18(1): 95-102.
Frewer, L., Howard, C., Hedderly, D., and Shepherd, R. (1996). “What determines Trust
in Information About Food-Related Risks? Underlying Psychological
Analysis. 16(4) 473-486.
Henson, S. (1996). Consumer Willingness to Pay for Reductions in the Risk of Food
Poisoning in the UK,Journal of Agricultural Economics 47(3) 403-420
Henson, S. and Northern, J. (2000). “Consumer Assessment of the Safety of Beef at the
Point of Purchase: A Pan-European Study.” Journal of Agricultural Economics,
Lobb, A., Mazzocchi, M., and Traill, W. (2007). “Modelling risk perception and trust in
food safety information within the theory of planned behaviour.” Food and
Quality Preference. 18: 384-395.
Lobb, A., Mazzocchi, M., and Traill, W. (2006). “ ‘Food Risk Communication and
consumers’ Trust in the Food Supply Chain:’ Report on socio-economic
determinants of trust in the food chain.” European Commission: The Fifth
Framework Programme 1998-2002. TRUST QLK1-CT-2002-02343.
Mazzocchi, M., Stefani, G., and Henson, S. (2004). “Consumer Welfare and the loss
induced by withheld information: the case of BSE in Italy.” Journal of
Agricultural Economics. 55(1):575-592.
McCluskey, J., Grimsrud, K., Ouchi, H. and Wahl, T., (2005). “Bovine Spongiform
Encephalopathy in Japan: Consumers’ Food Safety Perceptions and Willingness
to Pay For Tested Beef.” The Australian Journal of Agricultural and Resource
Economics. 49: 197-209.
McEachern, M., and Shroder, M. (2004) “Integrating the voice of the consumer within
the value chain: a focus on value-based labeling communications on the fresh
meat sector.” Journal of Consumer Marketing. 21:7 497-509.
Saghaian, S. and Reed, M. (2004). “Demand for Quality-differentiated Beef in Japan.”
German Journal of Agricultural Economics. 53(8): 344-351.
Taco Bell Website. <www.tacobell.com>
United States Department of Agriculture: Food Safety and Inspection Service. (2007).
Wade, M. and Conley, D. (1999). “Consumer Responses to Food Safety Information
from Print Media.” Paper presented at the World Food and Agriculture Congress:
International Food and Agribusiness Management Association, Florence, Italy,
June 13-16, 1999.
Table 1. Trust Component Factor Loadings for Respondents’ Trust of Food Safety Information from 19 Different Sources
Doctors/ health authority
Animal welfare organizations
Media (T4 )
*Values in bold are greater than or equal to .40 through Varimax Rotation.
Table 2. Descriptive Statistics
Number of People in Household
Age of Respondents
Average Weekly Chicken and/or Beef
Average Weekly Expenditure on
Chicken and/or Beef ($)
Table 3. Whom Respondents Trusted More Between the Respective Pairs
Concerning Food Safety Rumors
Family More than University Scientist
Family more than Public Authorities
Family more than Media
Family more than Producers
University Scientist more than Public Authorities
University Scientist more than media
University Scientist more than Producers
Public Authorities more than Media
Public Authorities more than Producers
Media more than Producers
Table 4. Percentages of Respondents’ Level of Agreement with Given Statements
Chicken and/or beef
Chicken and/or beef is
good value for money
Chicken and or beef is
not easy to prepare
Chicken and /or beef is
a safe food
Everyone in the family
likes chicken and/or
Chicken and/or beef
works well with lots of
chicken and/or beef is
low in fat
Chicken and/or beef is
low in cholesterol
Chicken and/or beef
Chicken and/or beef
helps the local farmers
I do not like the idea of
chickens and/or cows
being killed for food
Chicken and/or beef is
not produced taking
animal welfare into
Table 5. Parameter Estimates of ITP11 and ITP21
Trust in Suppliers (T1)
Trust in Gov’t/University
Trust in Organizations (T3)
Trust in Media (T4)
(*)-10% Significance level (**)-5% Significance level (***)-1% Significance level
Table 6. Parameter Estimates ITP22 Reporting only Statistically Significance At
10% Level or Greater of Demographic Shifters
(*)-10% Significance level (**)-5% Significance level (***)-1% Significance level [ ]-indicated influence
Figure 1. Pictorial Representation of SPARTA Model