1
CHAPTER 1. INTRODUCTION
This dissertation follows the non-traditional format and includes six
chapters. Chapter 1 is an overall introduction to this dissertation. Chapter 2
contains a comprehensive literature review for all research questions that guided
the dissertation. Chapters 3 - 5 include three articles corresponding to three
studies that will be submitted to research journals. Finally, Chapter 6 discusses the
overall findings related to all proposed questions, results, and conclusions. This
non-traditional format then follows with a comprehensive list of all references
used in Chapters 1-6 and any relevant appendix materials.
1.1 Introduction
Food safety is important as foodborne illness outbreaks cause great economic and societal
losses. According to the recent estimates by the Centers for Disease Control and
Prevention (CDC), each year one in six Americans (48 million) gets sick, 128,000 are
hospitalized and 3,000 lose their lives because of foodborne diseases (CDC, 2014a).
Foodborne illness outbreaks pose significant threats not only to public health, but to the
economy as well. Food Safety News estimates that foodborne illnesses cost the economy
more than 15.6 billion dollars each year (Flynn, 2014). At the restaurant level, foodborne
illness outbreaks are devastating as well. For example, Jack in
2
the Box – almost 30 years after their own E. coli outbreak – still suffers a stock price drop
whenever an E. coli outbreak happens. Often times, a company never fully recovers from
a major food safety problem (Seo, Jang, Almanza, Miao, & Behnke, 2014).
Efforts to protect public health and reduce foodborne illness outbreaks will not be fully
effective unless the resulting information is communicated to consumers. Improved food
safety communication can not only help consumers in making safer food choices and thus
reduce the instances of outbreaks, but can also provide motivation for businesses (e.g.
restaurants) to strengthen their food safety controls. In fact, studies have shown that
informing consumers about food safety will influence their behaviors (Choi, Nelson, &
Almanza, 2011; Porucznik & Royal DeLegge, 2013), and communicating food safety
information with consumers, such as publishing restaurant inspection scores, helps to
improve food safety controls (Almanza, Ismail, & Mills, 2002; Jin & Leslie, 2002).
However, food safety communications have not been particularly satisfactory (Worsfold,
2006). Traditionally, food safety information has been communicated through passive
medias such as TV and newspaper. Such communications have limited reach and
relevance and reduce its impact (Charles & Lawrence, 1990; Dutta-Bergman, 2004). If
food safety information were more accessible, consumers would be more likely to use it
(Worsfold, 2006). In this regard, the Internet presents great possibilities for
communicating food safety information to the public.
Internet-based platforms present a wide range of benefits that make them attractive
alternatives for food safety communications. But despite its potential, online food safety
communication has not always been successful (Bortree & Seltzer, 2009)(Thackeray,
3
Neiger, Smith, & Van Wagenen, 2012). This is partly because the media’s role has not
been studied extensively. When the lack of research is combined with consumers’
increasing interest in food safety (Food Safety News, 2016), the need to understand
media’s effect is pressing.
1.2 Research objectives
This research takes a user-centered approach, which utilizes user feedback and focuses on
user experience, and aims to understand the role of media in improving online food safety
communication. To achieve this overall objective, three progressive studies were
conducted. The first study (see Chapter 3) explored consumers’ communication
preferences and motivations, food safety information needs, and use of information. The
second study (see Chapter 4) employed Website Experience Analysis (WEA) to
understand how consumers experience websites used for food safety communication and
link website characteristics with users’ perceptions. The third study (see Chapter 5) built
upon the results from studies one and two in an extended EPPM model to test on a large
scale the impact of media characteristics on communication outcomes. Finally, best
practices were identified to help improve online food safety communication. See specific
research questions in Chapter 2. More specifically, the three studies are:
1. Food safety information on the Internet: Consumer media preferences.
2. How consumers interact with websites to obtain food safety information: An
application of Website Experience Analysis (WEA).
3. Improving food safety communication on the Internet: Influence of media on
different efficacy perceptions.
4
1.3 Research significance
The present research is an interdisciplinary research of food safety communication with a
combination of research methods, including a survey, Website Experience Analysis, and a
scenario-based survey. See Chapter 2 for more details. This research presented a
comprehensive view of consumers’ experiences using Internet-based platforms for
foodborne illness outbreak information as it incorporates consumers’ preferences with
actual observations of their behaviors (with a real website). Additionally, the results of the
three studies progressively extend the understanding of media’s role in food safety
communication. The first study explored preferences and underlying motivations and
priorities that lead to such preferences, the second study observed consumers’ interactions
with websites and linked evaluations of website characteristics to perceptions, and the
third study quantified the impact of media on behavioral intentions. The combined results
significantly advanced the knowledge of how media can be used to improve food safety
communication.
This research has both theoretical and practical applications. From the theoretical
perspective, this research presents the following contributions. First, this research extends
the scope of EPPM to incorporate novel constructs specific to media evaluations (process
and message efficacies), which is valuable as media is determined to be influential to
communication outcomes. This is the first time that process and message efficacies are
included, tested, and shown to be predictive of communication outcomes. This extension
may also support EPPM’s broader applications in communication and behavioral studies.
Second, the combination of three studies, utilizing a variety of theories and methods,
provides a useful and novel model for future studies of food safety communication. This
5
research demonstrated the power of utilizing complementary research methods to
examine the same core issue from different perspectives. Third, this research
demonstrates that website characteristics themselves are not the basis upon which
consumers form their intentions, but rather the meanings and perceptions users derive
from experiencing such characteristics are essential in determining behavioral intentions.
This highlights the need for future research to conduct user-centered research and shift
focus from specific website characteristics to concentrated efforts to understand
consumers’ experiences. Fourth, source, information, and media characteristics interact
and together they lead to perceptions and evaluations of the experience; going forward, it
may not be realistic or meaningful to consider the impact of such characteristics
separately. Fifth, the conceptualization of situational contexts for foodborne illness
outbreaks using susceptibility and severity is successful, providing a model for further
explorations. Sixth, food safety communication is unique and may be different from
other types of health communication in that consumers’ information needs are highly
personal and situational. This highlights the importance of further explorations of
consumers’ food safety information needs that incorporate contexts.
This research also supplies a number of practical applications. First, it was discovered
that foodborne illness communication is a two-step process. This suggests that to improve
communication effectiveness, communication efforts should focus on achieving different
goals at different stages of this communication process – more specifically, to broaden the
reach and inform more consumers at stage one, and to provide superior user experience at
stage two. Thus, different communication strategies (including media employed) are
appropriate. Second, the identified consumer preferences, particularly in media choices,
6
provide a guideline for resource allocations. Third, the identification of website
characteristics directly associated with consumers’ perceptions of information quality and
media usability, and later behavioral intentions, provide valuable design suggestions.
Fourth, the direct comparison of new media (social media) and traditional online media
(website) provides insights for entities that are attracted by social media’s potentials.
Fifth, some best practices were identified to help practitioners in managing their food
safety communication efforts. Sixth, the observations of actual consumer interactions
with existing websites highlight the areas for improvement and provides examples to
follow.
CHAPTER 2. LITERATURE REVIEW
In the non-traditional dissertation format, this chapter provides a
comprehensive discussion of all relevant literature for the topic of “Improving food
safety communication online: The role of media.”
2.1 Food safety and food safety communication
Food safety is essential in protecting public health and consumers can play an important
role in strengthening overall food safety control if they can get engaged and use food
safety information in their food decisions. The following sections discussed the
importance of food safety and food safety communication along with the current state of
food safety communication.
7
2.1.1 Food safety is important
Burden of foodborne illness. An estimated 48 million cases of foodborne illness occur
each year in the United States. In other words, one in six Americans gets sick from
foodrelated diseases every year. In addition, foodborne illness causes approximately
128,000 hospitalizations and 3,000 deaths. Eight known pathogens account for the vast
majority of illnesses, hospitalizations, and deaths; among these, norovirus causes the
most illnesses – 5.5 million each year. Although norovirus usually causes a mild illness, it
is also the fourth leading cause of death – 149 deaths each year – as it affects a very broad
range of population. Another well-known pathogen, nontyphoidal Salmonella, is the
largest cause of death and takes 378 lives each year (CDC, 2014a).
Historically, the U.S. (as well as other countries) has been burdened with foodborne
illness outbreaks. Some outbreaks have been so impactful that they have led to important
regulatory changes and continued to have lasting influences on the U.S. society. The case
of the Jack in the Box restaurant chain is one example. This outbreak happened in early
1993. At that time, 732 people were infected with a particularly virulent pathogenic strain
of E. coli found in undercooked beef patties in hamburgers. Many of them developed
serious complications, such as hemolytic uremic syndrome (HUS) that started with
bloody diarrhea (CDC, 1993). This case is extremely impactful for several reasons: 1) the
population most affected was children; 2) it generated widespread media coverage unlike
any other previous outbreak; 3) Jack in the Box was later identified in violation of a
recently passed cooking temperature law (state law); and 4) E. coli at the time was not
well known and people were not fully aware of its danger. As a result of this outbreak, E.
8
coli became a reportable disease to all state health departments and the Food and Drug
Administration (FDA) increased the cooking temperature for hamburgers from 140 °F
(60 °C) to 155 °F (68 °C). Jack in the Box never fully recovered from this negative event
– to this day, whenever an E. coli outbreak occurs, Jack in the Box is often mentioned in
media reports and their stock prices subsequently decline (Benedict, 2015; Seo et al.,
2014).
We have not been free of serious outbreaks in recent years. The CDC actively tracks the
number of cases each year, and a number of highly impactful and multi-state cases,
especially in recent years, suggest that the war against foodborne diseases is far from
over. In a 2012 outbreak and food recall related to Salmonella infections from cantaloupe,
261 cases were reported in 24 states; among the victims, 94 were hospitalized and 3
deaths were reported (CDC, 2012c). A massive recall – one of the largest food safety
recalls in the United States – happened in 2008-2009 when Salmonellacontaminated
peanuts and peanut-related products caused 714 cases of directly linked illness and may
have contributed to 9 deaths (CDC, 2013b). In another outbreak in 2012, 58 individuals
were reported to be infected with E. coli in nine states; of those, 33 (67%) were
hospitalized and 3 developed hemolytic uremic syndrome (HUS) (CDC, 2012b). A Foster
Farms chicken foodborne illness outbreak linked to Salmonella that started in 2013 and
spanned a 17-month period was finally declared over in July 2014. This outbreak led to a
total of 634 people becoming sick (the hospitalization rate was as high as 38%) in one of
the longest running outbreaks (CDC, 2014b). Another multistate outbreak of Salmonella
resulted in 356 victims in the same year (CDC, 2013a). In 2014, a Salmonella outbreak
linked to bean sprouts resulted in a total of 115 people becoming infected in 12 states. In
this outbreak, 25% of the ill required hospitalization (CDC, 2015d).
9
Additionally, a Listeriosis outbreak linked to Blue Bell creameries in 2015 resulted in 10
illnesses, 10 hospitalizations and 3 deaths (CDC, 2015c). The list goes on. In fact, the
CDC investigated 11 multi-state outbreaks in 2013, and 13 outbreaks in 2014, 11
outbreaks in 2015, and 7 so far (as of May) in 2016 (CDC, 2016a). In 2014, the worst 10
outbreaks where both the pathogen and the food source could be identified caused
illnesses for 1,356 people (Andrews, 2014).
Impact of food safety events. Food safety events take people’s lives, as seen in so many
outbreaks. Moreover, food safety events not only threaten the public’s health, they also
have significant economic costs. The economic unit at U.S. Department of Agriculture
(USDA) has data on costs associated with outbreaks linked to 15 major pathogens in the
U.S. These 15 pathogens together are responsible for more than 95% of the illnesses and
deaths from foodborne illnesses for which the CDC can identify the pathogen cause
(USDA, 2014). According to calculations done by Food Safety News, foodborne illnesses
cost the economy more than 15.6 billion dollars each year, and these foodborne diseases
cause more than 8.9 million people to become sick, 53,245 hospitalizations, and 2,377
deaths (Flynn, 2014).
Food safety events are not only threatening the public welfare, they often are devastating
to the companies associated with them. Foodborne illness outbreaks can generate recalls
and lead to reputation damages that can hardly be recovered, as in the case of Jack in the
Box. As a result, food safety events often lead to a lasting decline in product demand,
extended costs associated with the recall, and shrinkage in company value (Seo, Jang,
Miao, Almanza, & Behnke, 2013).
10
2.1.2 Food safety communication is important
The efforts to protect public health and reduce the number of people sickened in
foodborne illness outbreaks cannot be successful if food safety information is not
communicated to consumers. In fact, engaging consumers has a number of essential
benefits (Wilcock, Pun, Khanona, & Aung, 2004). Studies have shown that informing
consumers about food safety will influence their behaviors (Choi et al., 2011; Porucznik
& Royal DeLegge, 2013), and thus helps to improve food safety controls – for example,
improving inspection results (Almanza et al., 2002; Jin & Leslie, 2002) and reducing the
number of hospitalized cases (Jin & Leslie, 2002).
2.1.3 Effectiveness of food safety communication
Food safety communication is important and presents enormous benefits, but in the past,
it has not been particularly successful. Sometimes it is hard for consumers to obtain food
safety information, especially foodborne illness outbreak information. The following
sections discuss the effectiveness of food safety communication as well as possible ways
to improve such communication.
Traditional media. Traditionally, food safety information has been communicated through
push medias that are passive in message delivery, such as TV and newspapers. In fact, the
most common outlets for food safety information have been newspapers, television, and
radio (Almanza, Nelson, & Lee, 2003). The major drawback in push communication is
that users have few choices about what information they receive and when, so the
information is likely to have low relevancy, resulting in lower information usage.
11
Additionally, the limited reach of traditional food safety communications, combined with
low relevancy, reduces the impact of such communication (Charles &
Lawrence, 1990; Dutta-Bergman, 2004). If food safety information were more accessible and the
communication more engaging, consumers would be more likely to use it (Worsfold, 2006).
Internet-based food safety communication. In this regard, the Internet presents great
potential in improving food safety communications as it offers high speed, low cost, high
scalability, and high message fidelity (Glasgow, Klesges, Dzewaltowski, Estabrooks, &
Vogt, 2006; Snyder, 2001; Thackeray et al., 2012; Trouten, 2013). In fact, Internet
platforms have started to become the top choice for consumers for food safety
information (Bruhn, n.a.; Charanza & Naile, 2012). Along with consumers’ increasing
interest in food safety and demand for accountability and transparency through the food
system (Food Safety News, 2016), using the Internet for food safety communication
becomes more appealing. But on the other hand, unequal access and unfriendly and
unfamiliar designs hinder broader usage by the population at large (Cline & Haynes,
2001). It is thus even more important to examine how improvements can be made -
starting with understanding Internet-based communication.
2.1.4 Internet-based communication
Web 1.0 and 2.0. The World Wide Web (or web) – based on the Internet – has been
recognized to enhance human cognition and communication. The web was first introduced
by Tim Burners-Lee in 1989 (Kamel Boulos & Wheeler, 2007). Much progress has been
made to the web since its introduction, and the most important innovation might be its
evolving capabilities to support enhanced (e.g. fast and interactive) communications. To be
12
more specific, Web 1.0, which is the first generation of the web, could be considered merely
a means for disseminating/ broadcasting information. Web 1.0 applications allow limited
user interactions or content contributions, and are considered to be “read-only” (Aghaei,
Nematbakhsh, & Farsani, 2012). Unlike Web 1.0, Web 2.0 emphasizes users’ engagement
in the communication. It focuses on Internet-based media where users become producers
of information. User-generated content, usability, and interactivity are noticeable
characteristics of Web 2.0 (O'Reilly, 2005, 2009).
Social Media. Social media is a rather broad concept that represents the Web 2.0 mindset.
In general, social media is regarded as a service platform that engages users to
simultaneously support mass and personal communication (Ellison, 2007). In other
words, social media are Internet-based platforms that enable users to communicate with
each other, share ideas and information, and interact. In recent years, social media has
drawn scholarly attention as a promising alternative platform in health communication
(Morahan-Martin, 2004; Pew Research Center, 2015a; Sillence, Briggs, Harris, &
Fishwick, 2007; Xiao, Sharman, Rao, & Upadhyaya, 2014). Indeed, social media offers a
number of unique benefits such as potential high interactivity and good social support
that make it attractive for health communication (Cline & Haynes, 2001). But limited
research in the context of food safety communication makes the role and place of social
media largely unclear. Thus, to extend the understanding of media, social media is
directly compared with other more traditional Internet-based platforms, such as websites.
In this research, social media – characterized by interactivity and user-generated content –
is treated as a representation of the Web 2.0 concept and is compared to websites, which
traditionally have functioned as a Web 1.0 platform. Granted that there are discussions
13
over concepts beyond Web 2.0, such as Web 3.0 and Web 4.0 (Aghaei et al., 2012), they
are not as relevant in this discussion – as the comparison in this research focuses on the
platform’s communication features/characteristics, such as interactivity and the capacity
to support user-generated content. In sum, in this research, social media is used to
represent platforms that are interactive in nature and is directly compared with traditional
Internet-based platforms (e.g. websites) in food safety communications.
2.2 Information seeking behaviors
To understand consumers’ interaction with and usage of Internet-based platforms in an
attempt to improve food safety communication, it is important to understand how
consumers experience websites and seek information online.
2.2.1 Information needs – the drive
Consumers have different information behaviors driven by distinct information needs
(Case, 2012, pp. 68-69). Uncertainty reduction is a common information need (Henefer &
Fulton, 2005) that is particularly relevant to information behaviors that involve health
issues. Humans sense differences between what they know and what they want to know
and react by seeking information to reduce uncertainty (Atkin, 1973; Belkin, 1978;
Belkin, Oddy, & Brooks, 1982; Belkin & Vickery, 1985; Krikelas, 1983). In situations
where one’s health is at stake, it is not hard to imagine the need to reduce uncertainty by
seeking out information and reducing the gap between what is known and what needs to
be known. The assumption that humans would seek out information when facing
decisions regarding their health is rooted in psychological studies of human behaviors.
14
For example, in Maslow's hierarchy of needs, the most basic human needs are
physiological and safety needs – the needs for foods and safe foods; thus when facing
decisions, individuals seek information to reduce uncertainty and to meet their
physiological and safety needs (Maslow, 1943; Wilson, 1999).
2.2.2 Types of food safety information.
The FDA, CDC, and USDA, along with state and local governments (mainly health
departments) collectively protect public health by educating the public about food safety,
conducting restaurant inspections, and controlling and investigating foodborne illness
outbreaks (Almanza & Ghiselli, 2014; CDC, 2015a; FDA, 2015; USDA, 2015). These
efforts to ensure food safety result in different types of food safety information such as
restaurant inspection results, foodborne illness outbreak and recall information, and
seasonal food safety information. When developing appropriate communication
strategies, these distinct types of food safety information should be treated differently.
The primary reason for this distinction is that consumers may have different levels of
urgency and continuity in their information needs, and this will result in different
information seeking behaviors. For this reason, to yield meaningful results, this research
focuses only on foodborne illness outbreak information (FBI).
2.2.3 Components of communication
Without the understanding of how communication happens, any attempts at improving
communication outcome cannot be successful. The following sections review
components of communication.
15
Components of communication. Shannon and Weaver (1949) proposed a simplified
communication model known as the transmission model, suggesting that in order for
communication to occur, four components are necessary: an information source or sender
(which produces a message), a transmitter (which encodes the message into signals), a
media (from which signals are transmitted), and a receiver (which decodes or reconstructs
the message). Barnlund (1970) proposed a transactional model of communication and
incorporated interactions between senders and receivers of information. Additional
research has suggested the importance of purpose, context and timing (Agarwal, 2010).
These findings have been consistently verified (Fox, 1983; Shannon, 1949; Shrigley,
1978; Wilson, 2005). With these in mind, this research controls for context (FBI
outbreak) and timing (ongoing outbreak) while exploring the impacts of source, media,
and information on receivers’ behavioral intentions.
Distinction between source and media. In studying communications, the source (which
produces a message) and the media (through which signals are transmitted) should be
considered separately. This division is widely supported in the literature (Fox, 1983;
Shannon, 1949; Wilson, 2005). Moreover, this distinction is especially important in
Internet-mediated communication because source and media can have a different impact
on communication outcomes. For example, when one source utilizes multiple media to
communicate the same information, users may trust and thus use the information
communicated over one media more than another (e.g. a health professional uses a
personal blog vs. appearing on TV to promote the use of certain testing). Or people may
treat and use information gathered from the same media but from different sources
16
differently (e.g. news on the Internet from Fox News vs. CNN). The present research
focuses primarily on media effect in communication.
Interactions among source, media, and information. Uncertainties during a foodborne
illness outbreak would be expected to create a drive for information; thus motivated,
individuals’ evaluation and later usage of information would be impacted by information
quality, source characteristics, and media properties. When media can be evaluated
separately from the source, information/content characteristics interact with media
characteristics in determining communication outcome (Barry & Schamber, 1998;
Frighetto & Wolf, 2014; Krikelas, 1983; Kuhlthau, 1991; Park & Lessig, 1981; Tversky
& Kahneman, 1973; Young & Von Seggern, 2001). The characteristics of
communications (e.g. source, information, and media characteristics) may be experienced
differently by users and lead to different perceptions. For example, although all
consumers want to get accurate information, perceptions of information/content may vary
depending on the media which consumers use. It is thus important to examine users’
experiences and perceptions of such characteristics in the attempt to explore media’
impact on communication outcome.
2.2.4 Perceptions of website characteristics
Commonly, the characteristics of communications through websites (source, information,
and media) lead to users’ evaluations of information quality (how good the information
is) and media usability (how usable the media is) (Al-Qeisi, Dennis, Alamanos, &
Jayawardhena, 2014; Aladwani & Palvia, 2002; DeLone & McLean, 1992; Hanafi, Kasim, Ibrahim,
& Hancock, 2009; Huang & Benyoucef, 2013; Wixom & Todd, 2005). Dimensions that are most
17
commonly associated with evaluation of information quality and media usability were considered,
consolidated, and presented in Table 2-1. In sum, within the context of communicating foodborne
illness outbreak information, this research uses accuracy, timeliness, trustworthiness, applicability,
saliency, quantity, transparency, understandability, accessibility, interactivity, searchability, usability,
linkability, familiarity, and security to frame consumers’ priorities in media selection in the attempt
to understand consumers’ information seeking behaviors.
Table 2-1. Dimensions, definitions, subscales, and supporting literature for information quality and
media usability
Dimensions Definition
Subscales
Supporting literature
Information Quality
Accuracy
The information
is accurate
Accuracy
Ahn, Ryu, and Han (2007); Bailey and
Pearson (1983); Barry and Schamber
(1998); Jarke and Vassiliou (1997); Kim,
Eng, Deering, and Maxfield (1999); Lee,
Strong, Kahn, and Wang (2002); Liu and
Arnett (2000); McKinney, Yoon, and
Zahedi (2002); Wang and Strong (1996);
Wixom and Todd (2005); Young and
Von Seggern (2001); Zmud (1978)
Timeliness
The information
is current
Timeliness
Ahn et al. (2007); Aladwani and Palvia
(2002); Bailey and Pearson (1983); Barry
and Schamber (1998); Jacob, Lok,
Morley, and Powell (2011); Jarke and
Vassiliou (1997); Kim et al. (1999); Liu
and Arnett (2000); McKinney et al.
(2002); Wixom and Todd (2005); Young
and Von Seggern (2001); Zmud (1978)
Trustworthiness
The source of
information is
trustworthy
Trustworthiness
Carlson (2007); Goodhue and Thompson
(1995); Janneke de et al. (2004); Kim et
al. (1999); Marlow (2004); Pieniak,
Verbeke, Scholderer, Brunsø, and Olsen
(2007); Yang, Cai, Zhou, and Zhou
(2005)
Table 2-1 continued
18
Applicability
The information
is applicable
Applicability
Ahn et al. (2007); Aladwani and Palvia
(2002); DeLone and McLean (1992);
HernáNdez, JiméNez, and Martín (2009);
Larcker and Lessig (1980); Lin and Lu
(2000); McKinney et al. (2002)
Saliency
The information
is detailed and
specific
Detailed
information/
specificity
Aladwani and Palvia (2002); Bailey and
Pearson (1983); Barry and Schamber
(1998); DeLone and McLean (1992);
Kim et al. (1999); Knijnenburg et al.
(2013); Langfield-Smith (1997); Liu and
Arnett (2000); McKinney et al. (2002);
Wixom and Todd (2005); Yang et al.
(2005)
Quantity
There is a large
amount of
information
Information
quantity
Kim et al. (1999); Petts et al. (2000);
Yang et al. (2005); Zmud (1978)
Transparency
The source of
information is
visible
Transparency
Kim et al. (1999); Marlow (2004);
Pidgeon and Kasperson (2003); Rutsaert
et al. (2013); Seeger (2006)
Understandability
The information
is easy to
understand and
clear in meaning
Easy to
understand
DeLone and McLean (1992); King and
Epstein (1983); Lee et al. (2002);
McKinney et al. (2002)
Media usability
Accessibility
The access is fast
and easy
Speed
Aladwani and Palvia (2002); Bailey and
Pearson (1983); Huang and Benyoucef
(2013); Kim et al. (1999); Krikelas
(1983); Kuhlthau (1991); Lin and Lu
(2000); Liu and Arnett (2000); McKinney
et al. (2002); Wixom and Todd (2005)
Ease of use
Aladwani and Palvia (2002); Barry and
Schamber (1998); Krikelas (1983);
Kuhlthau (1991); Yang et al. (2005);
Young and Von Seggern (2001)
Interactivity
Users can
interact with
others, including
other users and
information
sources
Interactivity
Aladwani and Palvia (2002); Kim et al.
(1999); Kuang and Cho (2015); Palen,
Vieweg, Liu, and Hughes (2009);
Rutsaert et al. (2013); Shepherd (2008);
Sundar (2007); Verbeke (2011); Yang et
al. (2005)
19
Searchability
The navigation/
search is easy
Searchability
Aladwani and Palvia (2002); Alexander
(2014); Barry and Schamber (1998);
Papadomichelaki and Mentzas (2012);
Young and Von Seggern (2001)
Table 2-1 continued
Appearance
The website is
visually
appealing
Visuals – use of
pictures,
videos, and
appropriate
colors
Aladwani and Palvia (2002); Mei,
Bansal, and Pang (2010); Tarafdar and
Zhang (2005); Verbeke, Frewer,
Scholderer, and De Brabander (2007)
Linkability
The website
supports
outbound links
to additional
information
Linkability
Aladwani and Palvia (2002); Bailey and
Pearson (1983); Kim et al. (1999);
McKinney et al. (2002); Miller and
Doyle (1987); Nelson, Todd, and Wixom
(2005); Sethi (2007)
Security
The information
seeking is
conducted in a
secure manner
Privacy settings/
controls and
few
advertisements
Al-Qeisi et al. (2014); Aladwani and
Palvia (2002); Lee and Kozar (2006); Liu
and Arnett (2000); Min, Li, and Zhong
(2009); Papadomichelaki and Mentzas
(2012); Wang and Strong (1996)
No viruses
Freeman (2012); Freeman and Chapman
(2008)
Does not track
user data
Aladwani and Palvia (2002)
Familiarity
The website is
familiar to users/
users have used
the website
before
Familiarity
Frighetto and Wolf (2014); Park and
Lessig (1981); Tversky and Kahneman
(1973).
2.2.5 Perceptions of information quality and media usability
Characteristics of a website are experienced by users and contribute to formations of
different perceptions and eventually result in different communication outcomes or
behavioral intentions. Wixom and Todd (2005) made a significant contribution in
highlighting media’s significant role in communication by integrating previous literature in
user satisfaction and technology acceptance and proposing a model that clearly
20
distinguishes users’ perceptions and beliefs about the system (i.e. perceptions of
information quality and media usability) from users’ perceptions and beliefs about using
the system and its information. They further theorized that perceptions or beliefs towards
information quality and system quality (i.e. media usability) influence behavioral
intentions (Wixom & Todd, 2005). Thus, this research further theorizes that website
characteristics experienced by consumers will lead to perceptions towards source,
information, and media, thus resulting in perceptions of information quality and media
usability; and beliefs of information quality and media usability will eventually influence
behavioral intentions.
2.3 Theoretical framework
While website characteristics may influence consumers’ behavioral intentions, it is still
unclear how this influence can take place. In studying health communications,
particularly in situations where risk is involved, threat and efficacy perceptions are
powerful in understanding consumers’ behaviors. The models/ theories of Protection
Motivation Theory, Parallel Response Model, and the later Extended Parallel Processing
Model (EPPM) are widely cited as predictive of health and protection related behavioral
intentions.
2.3.1 Fear appeal
The models begin with a fear appeal. The idea of a fear appeal is to communicate the risk
so people are alarmed and paying attention, thereby increasing the likelihood of accepting
the communicated message and avoiding the risk conveyed (Hovland, Janis, & Kelley,
1953). Risk represents danger and the possibility of a negative outcome. The inherent
21
logic of fear appeal is that if a message is persuasive enough that it makes people aware
of how much danger they are putting themselves in by engaging in a certain behavior,
this awareness would be expected to result in behavior changes in order to avoid danger
and reduce risk (Yzer, Southwell, Stephenson, Rice, & Atkin, 2012). Four theories
predominate in the fear appeal domain of research. These are the fear-as-drive model, the
parallel process model, the protection motivation theory, and the extended parallel
processing model.
Fear was initially deemed to be a drive as suggested by Hovland et al. (1953). A drive
motivates people to change and the changes are considered responses. These
changes/responses are made to improve emotions (e.g. the fear) and not necessarily to
avoid the threat. Because of this, early attempts to induce fear to promote changes in
behaviors did not produce consistent results as people could deny the risks. So Hovland
et al. (1953) introduced the interpretation of “recommendation” in a fear appeal. In other
words, a message or appeal should not only include a fear-inducing component but
provide a coping strategy or recommendation as well. If the recommendation works to
avert the threat and thus reduce emotional distress, the recommended action will become
habitual when similar cues of threat are present. On the other hand, if the
recommendation does not work as anticipated to reduce the unpleasant emotional state,
people revert to more maladaptive strategies such as defensive avoidance (not thinking
about the threat) and perceived manipulation (feeling manipulated thus angry). Janis
(1967) later introduced the curvilinear model of fear – fear-as-drive model. In this model,
fear inducement up to a certain level will increase adaptive responses, but when fear
22
becomes too strong, adaptive responses will decrease. Thus, moderate fear is considered
most effective in motivating desired behavioral changes.
The Parallel Response Model was proposed by Leventhal and Watts (1966) to extend the
understanding of how individuals process threat and fear. This model proposed that fear
is not the only immediate response after a fear appeal. In this model, a threat is processed,
which then triggers danger control or fear control or both. Danger control occurs when
people process the threat and engage in actions to avert the threat while fear control
happens when people think about strategies to control their fear (not necessarily the
threat). This model does not stipulate when people engage in danger control, fear control,
or both and if these two processes interact. In response to the lack of specificity, Rogers
(1975) introduced and later improved (1983) the Protection Motivation Theory (PMT).
PMT focuses only on the danger control response side of the parallel response model.
PMT proposed that there are four components involved in a fear appeal: perceived
vulnerability, perceived severity, response efficacy, and self-efficacy. PMT further
suggests that these four components will produce two processes: threat appraisal and
coping appraisal. In their model, perceived vulnerability and perceived severity induce a
threat appraisal, while response efficacy and self-efficacy elicit a coping appraisal. The
threat appraisal assesses the severity of the situation and the coping appraisal frames how
people respond to the situation. Protection motivation (leading to desired attitude and
behavioral changes) will be maximized if both appraisals are positive. PMT advances our
understanding of risk communication in that it helps to define the nature of threat and
coping components of a fear appeal; however, it did not explore fear’s role as an emotion
in explaining and promoting behavioral changes.
23
2.3.2 The Extended Parallel Processing Model (EPPM)
EPPM. Witte (1992) reintegrated affective process with Rogers’s work of cognitive
process of fear (1975, 1983) and built upon Leventhal’s framework of parallel danger and
fear processes (1970). She introduced the Extended Parallel Process Model (EPPM).
EPPM suggests that after exposure to a fear appeal, an individual forms the perception of
threat based on the severity of the threat and how susceptible the individual is to the
threat. The more severe the situation and the more susceptible the individual is, the more
they are motivated to begin the second appraisal – evaluation of the efficacy of the
recommended response. Their perception of response efficacy (how effective is the
recommended response in reducing the threat) and self-efficacy (if they believe they can
perform the recommended response) is formed and combined with their perceived level
of threat, thereby leading to their response choice. There are three possible outcomes
after the fear appeal. Individuals can ignore the message if the threat is perceived to be
low or irrelevant (so the efficacy appraisal is not even triggered). Alternatively,
individuals can engage in danger control, where they process the message and take the
recommended actions. This is more likely to occur from a combination of high threat
perception and high efficacy perception. Or individuals can engage in fear control when
they are exposed to a serious threat without a perceived effective response or they do not
believe they can successfully perform the response and avoid the threat. In other words, a
high threat combined with low efficacy will trigger danger control. Fear control is an
emotional process, unlike danger control, which is processed cognitively. When engaged
in fear control, people cope and control their fear, but not the danger. See Figure 2-1.
24
Figure 2-1. Extended Parallel Process Model (EPPM)
Source: Witte (1994)
Why EPPM. EPPM’s quality has been evaluated (Maloney, Lapinski, & Witte, 2011) and
results of several meta-analyses suggested that EPPM is powerful and useful (Webb,
Joseph, Yardley, & Michie, 2010; Witte & Allen, 2000). EPPM is particularly applicable
to this research for the following reasons. First, the conceptualization of antecedents –
threat and efficacy of behavioral changes matches well with the reality of food safety
communication. The focus of food safety information in this research will be foodborne
illness outbreaks because these are primary examples where a timely need for food safety
information may manifest in an Internet search for information. A foodborne illness
outbreak is inherently risky and presents immediate hazard, so it can be conceptualized as
the fear stimuli that would trigger the perception of threat. Effective communication
through websites would be helpful in empowering audiences to make safer dining choices
thus improving their perceptions of efficacy, so the audiences will accept and use the
information and manage the threat (instead of ignoring the risks). In other words, EPPM
conceptualization of motivational factors (based on fear) that would activate behavioral
response can be applied in food safety communication. Second, EPPM, by incorporating
25
both threat and efficacy components, indicates how to improve communication outcomes.
More specifically, EPPM predicts that when confronted with fear-inducing stimuli, a
higher level of perceived efficacy will increase message acceptance (Witte, 1994). In this
regard, EMMP allows the identification of ways to improve outcomes of food safety
communication online. Third, EPPM has flexibility that allows the incorporation of
website characteristics (Witte, 2015). Websites present unique media characteristics (e.g.
interactivity and searchability) and content characteristics (e.g. the possibility of more
timely updates) that could change users’ perceptions of efficacy and/or threat.
Some recent critics of EPPM have argued that EPPM does not take into consideration the
possible interaction between threat and efficacy perceptions. A recent meta-analysis has
provided evidence that there is an interaction effect between threat and efficacy - threat
only had an effect under high efficacy and efficacy only had an effect under high threat
(Peters, Ruiter, & Kok, 2013). Because of this, this research only focuses on possible
improvements in communication outcomes when consumers are confronted with high
level of threat. Future studies may explore other possibilities.
2.3.3 A new take on EPPM
Focus of the prediction. Since its introduction, EPPM has been widely adopted to develop
effective risk communication messages, specifically messages that would elicit adaptive
behavioral responses (Gore & Bracken, 2005; McMahan, Witte, & Meyer,
1998). Outside message design, EPPM has been applied to assess attitudes towards an
action (Barnett et al., 2009) and predict behaviors (Hullett & Witte, 2001; Roberto &
Goodall, 2009). This research attempts to build upon and extend EPPM’s traditional
26
applications and test to determine if EPPM can be used to examine effects of media on
communication effectiveness.
Incorporates constructs specific to communication media. Health communications are
increasingly happening over digital media (Rice & Atkin, 2012). To incorporate
perceptions related to unique characteristics of websites, this research proposes two
additional efficacy components - message efficacy (perception of information quality)
and process efficacy (perception of media usability) to be included in the model. In this
research, the four types of efficacy are defined as the following: Self-efficacy - a person’s
belief that he or she has the ability to find needed information (about foods and
restaurants to avoid); response efficacy - a person’s belief that using the information
found reduces the risk of getting sick, message efficacy - a person’s belief that the
information found is of high quality; process efficacy – a person’s belief that the process
of finding needed information is easily done.
Construct operationalization. Research has suggested that the operationalization of
EPPM constructs lacks consistency (Popova, 2012), and this is also reflected in literature
review. So, instead of applying any pre-existing operational definitions of the constructs,
this research started with thoroughly developed theoretical concepts in EPPM and
operationalizes these theoretical concepts/ constructs into measureable variables
according to the context of this research. See Study 3 for more details.
2.4 Research questions
In the past, media’s potential contribution in improving communication outcomes has
received limited scholarly attention. Overall, within the context of foodborne illness
27
outbreak communication, the role of media has been largely overlooked (Rutsaert et al.,
2013). Researchers have focused on benefits of new media (Rutsaert et al., 2014), effects
of communication and intervention (Mayer & Harrison, 2012; Mitchell, Fraser, &
Bearon, 2007; Powell, Jacob, & Chapman, 2011) , user typology (Kuttschreuter et al.,
2014), and message formation (Gordon, 2003). This lack of research limits scholars’
ability to fully explore ways to improve usage of Internet-based communication platforms
in food safety communication. To address this issue, this research utilized three studies to
examine media’s role in food safety communication. The three studies are laid out as:
1. Food safety information on the Internet: Consumer media preferences – to explore
consumers’ media preferences and motivations, food safety information needs,
and use of information.
2. How consumers interact with websites to obtain food safety information: An
application of Website Experience Analysis (WEA) – to understand how
consumers experience websites used for food safety communication and link
website characteristics with users’ perceptions.
3. Improving food safety communication on the Internet: Influence of media on
communication outcome – to examine the impact of media characteristics and
different efficacy perceptions on communication outcome; and to identify best
practices when using Internet-based platforms for food safety communication.
More specifically, this research is guided by the following research questions:
1. What are the preferred media for consumers to obtain foodborne illness outbreak
information? (Study 1)
28
2. What are the priorities for consumers in their media selection? (Study 1)
3. What website characteristics are experienced by consumers that influence their
efficacy perceptions (response, self, process, and information) towards the
communication? (Study 2)
4. What are the relationships among the efficacy perceptions (self, response, process,
and message)? (Studies 2 and 3)
5. How will different Internet platform characteristics (media, information, and source
characteristics) contribute to improving food safety communication outcomes
(behavioral intentions)? (Study 3)
6. When confronted with a threat, will a higher level of perceived efficacy (including
process and message efficacy) improve communication outcomes? (Study 3)
7. Is EPPM useful in understanding food safety information-seeking behaviors? (Study
3)
The overall research framework of the three studies is shown below in Figure 2-2.
Figure 2-2. Research Framework
29
2.5 Research method choices
To address the research questions, a combination of three studies was designed and
implemented. The first study utilized a survey to gauge consumer preferences,
motivations, and needs. Building on the results of Study 1, Study 2 employed Website
Experience Analysis (WEA) and mapped website characteristics to consumer
perceptions. Finally, Study 3 tested how changes in media perceptions would impact
communication outcomes from Studies 1 and 2 in the extended EPPM model using a
scenario-based survey.
2.5.1 Review of method choices
There are three major streams of research that are relevant to studies about
communication effectiveness. Each stream relies more heavily on a specific type(s) of
methodology to gather and analyze data. For example, one research stream focuses on the
audience or information receiver and uses survey and interview methodologies. A second
focuses on the information source and message, or social media posts in this case. This
research stream utilizes content analysis, machine mining, or analysis of social media
posts. Finally, the third stream focuses on the communication process or interaction, as
well as the relationships among the information users and between the source and the
audience. This stream often conducts network analyses.
Because the purpose of Study 1 was to understand consumers’ motivations and
preferences, a survey methodology was utilized to gather both quantitative data and some
open-ended response data. For the second study, which mapped consumers’ perceptions
of website characteristics, Website Experience Analysis (WEA) was applied to collect
30
qualitative data. Finally, the third study, which addressed the relationships among website
characteristics and users’ perceptions and influence of perceptions on communication
outcomes, employed a scenario-based survey to collect quantitative data on a larger scale.
The reasons for these choices are discussed below.
2.5.2 Survey
Among the most widely used research methods in the social sciences, the survey has been
widely used to gauge participants’ perceptions, understanding, motivation and reasoning
for actions. Compared to interviews and focus groups, which are used to collect more
detailed and in-depth qualitative data, the survey is generally used for collecting
quantitative data on a large scale. The survey focuses on the audience in communication,
especially targeting a better understanding of the audience’s attitudes, perceptions,
motivations, etc. Surveys are also useful in establishing or testing relationships among
variables as it provides quantitative data to support statistical analysis (such as
regression). For example, Ellison, Steinfield, and Lampe (2007) used a survey to collect
quantitative data to examine the relationship between the use of Facebook and the
formation and maintenance of social capital. In another study, a survey was used to study
consumers’ motivations for reading and contributing to restaurant reviews on Yelp
(Parikh, Behnke, Vorvoreanu, Almanza, & Nelson, 2014). Thus, surveys were considered
to be appropriate for Study 1 and 3 data collection.
2.5.3 Website Experience Analysis
Rather than characteristics of a website, the second study focuses on users' perceptions of
such characteristics. It is users’ perceptions towards website characteristics rather than
31
website characteristics that influence users’ behavioral intentions. Website Experience
Analysis (WEA) allows users’ experiences and perceptions of the websites to be captured
and thus was considered as the research method for Study 2.
Website Experience Analysis (WEA) was proposed by Mihaela Vorvoreanu as a new
research protocol in studying website experience (Vorvoreanu, 2004, 2006, 2008a,
2008b). WEA adopts a research perspective that focuses on the user’s experience rather
than the structural characteristics of the platform alone. The reasons for adopting a
usercentered approach are grounded in communication theories. Meaning is not
embedded in the text, waiting to be discovered by users. Meaning is created in the
process of reading and interacting with the website (Fish, 1980). Thus, methodologically,
analyses based on texts (such as content analysis and analysis of social media posts)
should be replaced by observation and analysis of people’s experiences and perception or
attitude changes while using and interacting with the media (Vorvoreanu, 2004).
WEA is particularly suitable for answering the research questions of Study 2 as it allows
mapping of users’ perception to specific characteristics of the media. Additionally, WEA
supports the mapping of users’ perceptions to specific characteristics of the media, which
extends the understandings of users’ perceptions that were not possible to capture in
Study 1 and builds foundations for Study 3. For example, Study 1 found that timeliness is
important, but it was unclear if timeliness is directly related to evaluation of the media or
the information. As previous literature provides inconsistent results (Aladwani & Palvia,
2002; Wixom & Todd, 2005), it is necessary to map website characteristics to consumers’
perceptions before such characteristics’ role in influencing communication outcomes can
be examined.
32
In addition, WEA offers a few unique benefits. First, WEA allows direct observations to
be made of users’ behaviors. This overcomes the potential biases presented by survey and
interview research (e.g. what users think they would do may be different from what they
actually would do). Second, WEA has built-in flexibility that allows its applications in
accordance with different theoretical perspectives and measurement models.
The research methods do not have to be applied alone. In fact, a combined method can
often provide a more comprehensive understanding of the reality, as the understandings
gained from using different methods are often complementary (Goggins, Mascaro, &
Mascaro, 2012; Pak & Paroubek, 2010). As such, this research utilized different research
methods, as appropriate, for the three studies.
33
CHAPTER 3. STUDY 1: FOOD SAFETY INFORMATION ON THE INTERNET:
CONSUMER MEDIA PREFERENCES
The following study will be submitted to a peer-reviewed journal and is
included in this non-traditional thesis/dissertation as chapter 3. This article is
written in the APA style format.
3.1 Abstract
Foodborne illness outbreaks can cause considerable losses to the economy and society. In
the past, efforts in communicating food safety information to consumers have not always
been successful, partly because user expectations and preferences were not fully explored
and understood. This study uses an online questionnaire to explore consumer media
preferences in food safety communication, particularly about communication of
foodborne illness outbreak information. Results of this study demonstrate the importance
of the Internet as a media in communicating information about foodborne illness
outbreaks. Despite the great potential of social media, websites are the preferred online
platform for communicating foodborne illness outbreak information. Media
characteristics, together with information and source characteristics, influence
consumers’ evaluation and preference of a media. Information quality is the most
important priority for consumers. Aside from it, searchability is the most valued media
functionality for consumers when looking for foodborne illness outbreak information.
34
3.2 Introduction
Food safety is extremely important, as foodborne illness outbreaks can cause considerable
losses to our economy and society (CDC, 2014a; Flynn, 2014). Foodborne illness
outbreaks can be devastating to a company or brand as well. For example, Jack in the
Box –almost 30 years after their own outbreak – still suffers a stock price drop whenever
an E.coli outbreak occurs (Seo et al., 2014). Oftentimes, a company never fully recovers
from a major food safety event (Seo et al., 2014). But efforts in ensuring food safety
would not be successful if the resulting information is not communicated to consumers.
In fact, communicating food safety information has been shown to be effective in
engaging consumers (Choi et al., 2011; Porucznik & Royal DeLegge, 2013) and
improving food safety controls – for example, improving inspection results (Almanza et
al., 2002; Jin & Leslie, 2002) and reducing the number of individuals hospitalized (Jin &
Leslie, 2002).
When evaluating communication efforts, it is important to separate media effect from
source effect (Fox, 1983; Shannon, 1949; Shrigley, 1978; Wilson, 2005). Communication
involves a number of elements, and the source (from whom) and the media (how) should
be considered separately (Shrigley, 1978). This distinction is especially important in
Internet-mediated communication because source and media can have different impacts
on communication outcomes. For example, when one source uses multiple medias to
communicate the same information, users may trust and thus use the information
communicated over one media more than another (e.g. a health professional uses a
personal blog vs. appearing on television). Or, people may view information differently if
it was gathered from the same media but from different sources (e.g. news on the Internet
35
from Fox News vs. CNN). Overall, within the context of foodborne illness outbreak
communication, the role of media has been largely overlooked (Chapman, Raymond, &
Powell, 2014). Because of the importance of the issue, this study utilizes user feedback
and explores consumer media preference, especially online media preference, for
foodborne illness outbreak information.
Past food safety communication efforts have not always been successful as consumers
feel it is sometimes hard to obtain food safety information, especially foodborne illness
outbreak information (Worsfold, 2006). When it comes to communications, there are
generally two forms, push and pull communication. Generally speaking, it is considered
“pull” if users request and retrieve the information; and it is considered “push” if the
information is sent in anticipation of users’ needs or in other words, the information is not
directly solicited (Cybenko & Brewington, 1999). Traditionally, food safety information,
along with other health-related information, has been communicated through push medias
such as TV and newspapers. The major drawback of push communication is that users
have few choices about what information they receive and when. This has limited the
reach and relevance of food safety information communicated and thus the impact
(Charles & Lawrence, 1990; Dutta-Bergman, 2004). Consumers indicated that if food
safety information were more accessible, they would be more likely to use it (Worsfold,
2006). In this regard, the Internet, especially emerging platforms such as social media
sites, presents great possibilities for communicating food safety information to the public.
Internet-based platforms, particularly social media sites, have the potential to be
appropriate and effective for food safety communication since these platforms enjoy the
benefits of timeliness (Tinker & Fouse, 2009), high accessibility (Duggan, Ellison,
36
Lampe, Lenhart, & Madden, 2015), improved usability – e.g. multi-media – (Fischhoff,
2012), cost effectiveness (Thackeray et al., 2012; Trouten, 2013), high scalability
(Glasgow et al., 2006), and high message fidelity (Snyder, 2001). In addition, social
media provides some unique advantages that would potentially help to improve food
safety communication. First, social media is widely adopted and has very high user
engagement, especially among younger groups, and increasingly among older consumers
as well (Pew Research Center, 2015b, 2015c). Thus, social media may help food safety
communication reach a broader audience, including previously hard-to-reach populations
(Brenner, 2013; Chou, Hunt, Beckjord, Moser, & Hesse, 2009; George Ettel Iii, Ettel,
Wilson, & Meola, 2012; O'Keeffe & Clarke-Pearson, 2011; Ramanadhan, Mendez, Rao,
& Viswanath, 2013). Second, social media is widely used in health and risk
communication (CDC, 2012a; Pew Research Center, 2015a) and has proven to be useful
(Tinker & Fouse, 2009) and powerful in prompting changes in users’ health behaviors
(Mou & Lin, 2014; Wantland, Portillo, Holzemer, Slaughter, & McGhee, 2004; Webb et
al., 2010). Third, social media has great potential in improving transparency, engaging
audiences, and building relationships because of its interactive and participatory nature
(Bekkers & Homburg, 2005; Chun, Shulman, Sandoval, & Hovy, 2010; Vorvoreanu,
2004, 2006, 2008a).
Recognizing the potential, efforts to use social media have been increasing (Harris,
Mueller, & Snider, 2013; Thackeray et al., 2012), but they have not always been fruitful
(Ma, Almanza, & Ghiselli, 2015). This is partly because consumer preference and
expectation, especially regarding media, are not clearly understood. In understanding
consumer media preferences, it is perhaps more valuable to explore the reasons for the
37
preferences. To study consumers’ priorities in evaluating a media, it is important to note
that characteristics of other communication components such as information (e.g.
timeliness) and source characteristics (e.g. trustworthiness) can all play important roles in
consumer evaluation and later selection of a media (Al-Qeisi et al., 2014; Aladwani &
Palvia, 2002; Barry & Schamber, 1998; DeLone & McLean, 1992; Frighetto & Wolf,
2014; Hanafi et al., 2009; Huang & Benyoucef, 2013; Krikelas, 1983; Kuhlthau, 1991;
Park & Lessig, 1981; Tversky & Kahneman, 1973; Wixom & Todd, 2005; Young & Von
Seggern, 2001). Thus, this study incorporated the most commonly cited media,
information, and source characteristics (see Chapter 2 for a complete list) and adapted
them to the context of foodborne illness outbreak communication to frame consumers’
priorities in media preference and selection. These characteristics include: accuracy,
timeliness, trustworthiness, applicability/saliency, interactivity, searchability, usability,
linkability, familiarity, and security. The following research questions were examined.
1. What is the preferred media for consumers to obtaining foodborne illness outbreak
information?
2. What are the priorities for consumers in their media selection?
3.3 Materials and Methods
An online questionnaire was used to capture consumers’ responses in five areas related to
foodborne illness outbreak information: 1) current usage of information, 2) preferred
media type to get information, 3) preferred Internet platform for information, 4) priorities
in media selection, 5) demographic information.
38
A foodborne illness outbreak scenario was used to frame the questions. The participants
were given a scenario in which there was a foodborne illness outbreak in their area, a
number of people became sick and many of those were hospitalized. Additionally,
information such as suspected foods was given to strengthen relevancy and make the
scenario more believable (“likely foods were thought to be chicken, lettuce, ground beef,
or possibly dairy products, including ice cream”). Then participants were asked to
evaluate different communication media options (e.g. Internet and TV) and Internetbased
platform options (e.g. social media and websites) in the process of selecting a restaurant
to visit.
Content experts with the communication, hospitality industry, and health inspection
experts were utilized to develop the questions and response choices. The questionnaire
was revised, refined, and finalized after two rounds of pilot testing. Text entries were
allowed to provide the opportunity for additional responses. Upon IRB approval, the
researchers distributed the survey to U.S. participants through an online company
(mTurk). This provided access to U.S. consumers across the nation. On average, the
survey took each participant 15 minutes to complete (upon completion, each participant
received a $0.50 payment). A total of 405 responses were collected in January 2016. If
responses were from the same IP address, used the same mTurk worker ID, were from
outside the U.S., used repeating mTurk code, or completed the questionnaire in less than
5 minutes, they were excluded from the analyses. After cleaning, a total of 370 usable
surveys were analyzed using SPSS version 23. Not all respondents answered all
questions.
39
3.4 Results
The demographic characteristics of the participants were summarized in Table 3-1. In this
study, there were more females (58.0%) than males (41.9%). A little more than half
(58.8%) of the participants were between the ages of 18-39, which is greater than the
U.S. population as a whole as indicated in the 2012 census data (given in parentheses in
Table 3-1). This might be because the data were collected online and younger populations
are more likely to be online. Because of the age distribution of the participants, more than
half of the participants (58.7%) did not have children. Regarding education, 47% of the
participants had less than a bachelor’s degree, 35.2% had a bachelor’s degree, and 17.8%
had degrees higher than a bachelor’s. Overall, the respondents had more education than
the population as a whole. Participants’ residential areas were comparable with the census
data.
Table 3-1. Profiles of Respondents (n=370)
Characteristics
n
%
Characteristics
n
%
Gender
Education
Male
118
41.9 (49.2)
Less than Bachelor’s Degree
132
47.0 (70.7)
Female
163
58.0 (50.2)
Bachelor’s Degree
99
35.2 (18.9)
Higher than Bachelor’s Degree
50
17.8 (10.4)
Age
18 – 29
81
28.7 (18.9)
Residential Area
30 – 39
85
30.1 (17.8)
New England
10
3.6 (4.7)
40 – 49
50
17.7 (19.3)
Mid Atlantic
42
15.3 (13.3)
50 – 59
45
16.0 (18.6)
East North Central
50
18.2 (15.2)
60 +
21
7.4 (25.3)
West North Central
22
8.0 (6.6)
South Atlantic
63
22.9 (19.3)
Household with Children
East South Central
11
4.0 (6.0)
No Children
165
58.7
West South Central
29
10.5 (11.7)
Children under 6 years old
43
15.3
Mountain
11
4.0 (7.2)
40
Children 6 years old and over
63
22.4
Pacific
37
13.5 (16.1)
Others
10
3.6
Note: numbers in parentheses are 2012 U.S. census data
3.4.1 Media preference
The Internet is the preferred media for foodborne illness outbreak information (see Table
3-2). To compare consumers’ overall preference for each media type, a “preference score”
was calculated. Scores were assigned for choices – the media would get 5 points every
time someone picked it as their number one choice, 4 points for a second choice, 3 points
for a third, 2 points for a fourth, and 1 point for a fifth choice – and then averaged using
the number of respondents. Results showed that the Internet received an overall score of
3.62, highest among all medias.
Table 3-2. Consumer media preference based on calculated preference score
Media Specific Internet-based platform Preference score
Internet 3.62
Websites 3.27
Facebook 2.78
Twitter 2.09
Instagram 1.76
TV 3.13
Face to face 2.92
Newspaper, book, magazine, or other printed material 2.81
Phone call or texts 2.52
41
Note: Assigning 4 points every time someone picked that media (or more specifically
platform) as their number one choice, 3 points for a second choice, 2 points for a
third, and 1 points for a fourth choice, after which an average was created by
dividing by the number of respondents.
Interestingly, while the Internet was the most preferred media for a great number of
participants (43%), some consumers ranked it as their least preferred choice (13%).
While having no noticeable difference in demographic characteristics (insignificant
ttests), this group of consumers was particularly concerned about information accuracy
and trustworthiness; they felt it was sometimes hard to determine the accuracy and
trustworthiness of information communicated over the Internet and preferred face-to-face
exchanges.
More specifically, among Internet-based platforms, consumers prefer to use websites to
find foodborne illness outbreak information, as 55% ranked websites as their most
preferred choice. Among social media sites, Facebook was ranked highest, with 24%
indicating it as their most preferred choice. Although Twitter is considered a platform for
news releases and timely updates (Chen, 2011; Java, Song, Finin, & Tseng, 2007), it was
least often selected as the respondents’ first choice (4%). These results were also
observed in the overall preference score calculation. Websites ranked the highest (3.27)
followed by Facebook (2.78).
3.4.2 Priorities in media selection
To better understand reasons behind consumer media preferences, participants were asked
for their priorities in media selection. ANOVA and post-hoc Tukey comparison results
suggest that that criteria used to evaluate Internet platforms/sites carried different weights
42
(F=186.254, p<0.001). Media, information, and source characteristics were all important
in consumer evaluation and selection of a media. More specifically, information quality
(accuracy and timeliness) carried the most weight when evaluating an Internet-based
platform. The most important media characteristic was that the platform had a search
function. See Table 3-3.
Table 3-3. Consumer priorities in media selection for foodborne illness outbreak
information
Priorities
N
Mean
SD
Accuracy
315
6.57a
0.97
Timeliness
314
6.27ab
1.04
Searchability (e.g. search function)
315
6.00bc
1.33
Security - less likely to have a virus
315
5.70cd
1.52
Trustworthiness (source)
314
5.57de
1.53
Linkability (e.g. links to additional information)
313
5.48de
1.42
Security - allows control of privacy setting
313
5.34def
1.61
Interactivity
315
5.30def
1.67
Applicability/Saliency
345
5.22def
1.63
Security - less likely to track user data
314
5.20ef
1.63
Security - fewer or no advertisements
315
5.19ef
1.60
Enhanced usability - visuals (e.g. pictures and videos)
314
5.02f
1.68
Familiarity
344
4.88g
1.75
Note: Mean is rated by participants on a 7-point Likert scale where 1= not at all
important and 7= extremely important.
*Based on multiple group comparison results (Tukey comparisons at α = 0.05)
The results suggest that social media may have suffered from mismatch between its
offerings and consumers’ priorities in selection a media when seeking foodborne illness
outbreak information. Social media supports more personalization and interactions, which
are considered to be less important, while not providing superior search function, which
is considered to be the most important media characteristic. Additionally, social media
may also suffer from its limited control over design features. For example, search
43
functions on social media sites are controlled by the social media owners or corporate
owners of the sites, so it is very difficult if not impossible to improve the feature.
In fact, many participants commented on their concerns about using social media in
gathering food safety information. The most commonly expressed concern was
information accuracy. The results were echoed when social media and websites were
directly compared using the performance criteria. See Table 3-4. Respondents’
perceptions were that the websites performed better for all criteria except one
(interactivity). Participants felt that websites, as compared to social media, supported
better searchability, provided more timely and accurate information, and supplied more
security. And websites particularly stand out as providing more accurate and trustworthy
information (78% and 76% more participants indicated that websites outperform social
media against the criteria of accuracy and trustworthiness). Social media’s major
advantage, interactivity, was not rated as important as the other priorities.
Table
1
-4. Performance comparisons (social media vs. websites) for foodborne illness
outbreak information against consumers’ priorities in media selection
1
.5.1 Conclusions and implications
Broader food safety communication may help engage consumers and contribute to an
overall improvement in food safety. Results of this study demonstrate the importance of the
Internet in food safety communications, particularly related to foodborne illness outbreaks.
44
Platform
Total
Criteria
Websites
Social
media*
Respons
es
Accuracy
89%
11%
329
Timeliness
57%
43%
330
Searchability
71%
29%
331
Security
54%**
46%**
331
Trustworthiness
88%
12%
329
Linkability
51%
49%
331
Interactivity
30%
70%
331
Applicability/Saliency
50%
50%
328
Usability
53%
47%
331
Familiarity
54%
46%
329
Note: the percentages represent proportion of participants indicated that website
or social media deliver superior performance against each criteria.
* e.g. Facebook, Twitter & Instagram.
** Average of the subcategories of security.
3.5 Discussion
Despite the great potential of social media, at the present time, website is the preferred
platform for communication about foodborne illness outbreaks. This does not mean social
media’s advantages are not valuable; its potential can be further explored in the future.
Consumer preference for website over social media for foodborne illness outbreak
information could be due to a number of reasons. First, social media’s potential for
45
informational use may not be fully recognized by consumers. Consumers currently
regard social media as a tool for socializing (2, 18). Furthermore, despite the growth of
social media adoption (for example, a number of health departments, such as the New
York State health department, are now using Facebook), social media has not been widely
used in food safety communications (10, 26). As a result, consumers may not even be
aware that food safety information can be obtained through social media. Second, aside
from providing superior performance against 9 out of the 10 important criteria consumers
use to evaluate and select a media, websites also appear to better match consumers’
priorities in media selection when looking for foodborne illness outbreak information. In
fact, 14% to 78% of participants indicated that websites perform better in the three most
important priorities consumers have when selecting a media for food safety information.
For example, consumers place a high value on a good search function, and websites
generally provide better search capabilities (including the use of indexing). Social media,
on the other hand, provides superior functionalities in areas that consumers believe to be
less important. For example, social media allows great interaction (asking questions and
getting responses), while consumers do not particularly value such interaction in their
search for foodborne illness outbreak information.
Additionally, social media’s support for user-generated content appears to be less valued.
Consumers prefer more accurate and trustworthy information from more “legitimate”
sources such as government or news companies rather than other users. Third, social
media may be overwhelming when trying to quickly locate needed information. In
support of these results, Robert and Dennis (39) have suggested that social media can
hinder the motivation and the ability to process the information communicated – e.g.
because of high social presence. Fourth, websites may appeal to consumers, as websites
46
are generally perceived to be less likely to track user data; for example, users do not need
to “log in” to view information.
Additionally, results of this study suggest that consumers not only prefer to use websites,
but they are, to a certain extent, against the use of social media. Consumers indicated that
they are not likely to use social media to obtain foodborne illness outbreak information.
This may be due largely to social media’s interactive and participative (user-generated
content) nature. While this allows collective intelligence to be harnessed, information
quality (e.g. accuracy and trustworthiness) in communication over social media cannot be
guaranteed. With a pressing need to obtain the most accurate, timely, and specific
information, consumers prefer to use a website for their information collection. Thus,
despite social media’s other benefits (speed, ease of use, and ability to get more
personalized information), it is not the preferred information media. It is possible that
social media may be more suitable as a supplement to websites. For example, social
media could be used to send updates with links to websites with the most current
information – so social media could serve as a “shortcut” that helps to direct consumers
to websites with more detailed information.
Consumer preferences for websites may, in fact, be good news in that websites allow
more controls over feature availability (e.g. search function) while social media features
are largely designed by the holding company (e.g. Facebook designs the look and the
search functionality on a Facebook page). Entities interested in communicating foodborne
illness outbreak information may better match consumer preferences using their websites
(rather than social media) and potentially have better communication results.
47
The results of this study have important implications for resource allocation. Agencies
that are interested in communicating food safety information, particularly foodborne
illness outbreak information, may want to focus their efforts on improving users’ website
experiences. Most importantly, providing a better search experience to quickly and
effortlessly locate needed information should become a priority. Providing an advanced
search function that allows targeted searches (e.g. sort by dates, locations, food sources,
types of illness, etc.) may be helpful.
Websites, if used appropriately, could provide some functions found on social media sites
– especially interactivity. Consumer preferences for more interaction might include:
places to offer feedback, sign up for newsletters (which also would increase engagement),
and follow updates on RSS feed (which would also build relationships).
Social media may still have potential in food safety communication, but further
exploration is needed. Additional research is needed to determine if social media should
be used purely as an extension of websites – e.g. posting links to websites – or whether
social media could enhance food safety communication beyond what can be
accomplished by using websites. From this study, it appears future scholarly efforts may
want to focus on studying Facebook’s potential in food safety communication. Twitter,
despite being argued as a feasible alternative, was not the preferred platform in this study.
The reason for this may be that Twitter has a maximum of 140 words and does not allow
detailed information (links are almost always used).
Information quality is highly valued by consumers. This shows that providing highquality
information (information that is accurate, up-to-date, and trustworthy) should still be the
48
most important priority in communicating foodborne illness outbreak information. When
information quality is maintained, consumers are flexible about where they get food
safety information. This highlights the possibility for entities interested in communicating
food safety information to utilize Internet-based platforms, thus reducing costs and saving
resources. But this study also shows that a small percentage of consumers are against the
use of Internet in seeking foodborne illness outbreak information. This group is highly
concerned about information accuracy and trustworthiness, and the Internet presents
challenges for them in evaluating the accuracy and trustworthiness of information
communicated. This indicates that, although powerful, the Internet may not be considered
a complete replacement to traditional media medias.
Agencies may want to use the Internet as a supplement or extension to traditional medias.
Additionally, it is important to note that media, information, and source characteristics all
influence consumers’ usage of information. This shows that while the communication
media has a great impact on communication outcomes, simply improving media
functionalities probably will not yield the best results. Information quality has to be
maintained or improved before better media functionalities can produce improved
communication outcomes.
Moreover, it appears that the line between source and media is somewhat blurred in the
minds of consumers (news sites, for example, appeared to be viewed as both the source
and the media type). To a certain degree, there does appear to be some inseparability of
the media and source on the Internet (as compared to traditional news sources that
included more print media). Particularly, if the source mainly communicates over
49
Internet-based platforms (for example, msn news [source] on the msn.com website
[media]), it becomes even harder for participants to make the distinction between source
and media. Entities interested in leveraging the Internet should keep this in mind and be
aware that, especially in online communication, the reputation of the information source
will influence the media usage, and the ways medias are used will reflect back (positively
or negatively) on the reputation of the information source.
Lastly, it appears that it is not necessary to use separate platform communication
strategies for different demographic groups, except for the group that is somewhat against
the usage of Internet (13% of the participants). While previous studies have suggested the
existence of potential individual differences, this study found that individual
characteristics had no significant impact on platform preferences (social media vs.
website). This may have occurred because social media and websites are both
Internetbased platforms. This finding suggests that the use of websites might be sufficient
in contacting hard-to-reach populations, such as younger consumers, with food safety
information (10). On the other hand, the existence of the group (13% of the participants)
that would not want to use the Internet for food safety information indicates the need to
maintain traditional medias for food safety communication; the Internet may not be a
preferred platform for certain populations and may be better used as a supplement to
other forms of communication.
3.5.2 Limitations
This study is not without limitations. In distributing the survey, the goal was to get a
representative sample of the U.S. population, but as seen, the participants are
50
concentrated in the younger age groups. Further, because the data was collected online,
the response group may have had more access to the Internet. Additionally, it is possible
that consumers, instead of evaluating source and media together, misunderstood the terms
and thus did not distinguish between source and media; further investigations are needed.
3.5.3 Future research
First, while the Internet is clearly the most preferred media, the possibility of employing
multiple medias should be explored. Additionally, as website was discovered to be the
preferred Internet platform, investigation is recommended into how consumers form their
perceptions towards a website; for example, what makes consumers perceive information
as being accurate and trustworthy and what makes consumers feel the platform is easy to
use. In this regard, a qualitative study that observes consumers’ actual interactions with a
website would be particularly insightful.
Further, it would be valuable to dive deeper into consumers’ reasons for their current
preferences. For example, why do consumers feel that websites outperform social media
against the criteria found to be important to their food safety information seeking? Is it
because the current food safety communication primarily happens on websites, so
consumers feel social media would not suit their needs in seeking for information? Or is
it because websites provide superior functionalities? Or is it because the reputations of
entities using various medias are different?
Lastly, it seems information, source, and media characteristics are, to a certain extent,
inseparable in consumers’ evaluation and selection of a media, so it would be insightful to
examine the relationships among these characteristics and determine how they interact
51
and influence communication outcomes together.
52
3.6 References
Al-Qeisi, K., Dennis, C., Alamanos, E., & Jayawardhena, C. (2014). Website design
quality and usage behavior: Unified theory of acceptance and use of technology.
Journal of Business Research, 67(11), 2282-2290.
Aladwani, A. M., & Palvia, P. C. (2002). Developing and validating an instrument for
measuring user-perceived web quality. Information & Management, 39(6),
467476.
Almanza, B. A., Ismail, J., & Mills, J. E. (2002). The impact of publishing foodservice
inspection scores. Journal of Foodservice Business Research, 5(4), 45-62.
Barry, C. L., & Schamber, L. (1998). Users' criteria for relevance evaluation: A
crosssituational comparison. Information Processing & Management, 34(2–3),
219236.
Bekkers, V., & Homburg, V. (2005). Information ecology of e-government: Egovernment
as institutional and technological innovation in public administration
(informatization developments and the public sector). Amsterdam, Netherlands:
IOS Press.
Brenner, J. (2013). The demographics of social media users - 2012. Retrieved from
http://www.pewinternet.org/files/oldmedia/Files/Reports/2013/PIP_SocialMediaU
sers.pdf
53
CDC. (2011). Estimates of foodborne illness in the United States. Retrieved from
https://www.cdc.gov/foodborneburden/
CDC. (2012). The health communicator’s social media toolkit. Retrieved from
http://www.cdc.gov/healthcommunication/toolstemplates/socialmediatoolkit_bm.
pdf
Chapman, B., Raymond, B., & Powell, D. (2014). Potential of social media as a tool to
combat foodborne illness. Perspectives in Public Health, 134(4), 225-230.
Charles, K. A., & Lawrence, W. (1990). Mass communication and public health:
Complexities and conflicts. Newbury Park, CA: Sage.
Chen, G. M. (2011). Tweet this: A uses and gratifications perspective on how active
Twitter use gratifies a need to connect with others. Computers in Human
Behavior, 27(2), 755-762.
Choi, J., Nelson, D. C., & Almanza, B. (2011). The impact of inspection reports on
consumer behavior: A pilot study. Food Control, 22(6), 862-868. Retrieved from
http://www.sciencedirect.com/science/article/pii/S0956713510003634
Chou, W.-Y. S., Hunt, Y. M., Beckjord, E. B., Moser, R. P., & Hesse, B. W. (2009).
Social media use in the United States: Implications for health communication.
Journal of Medical Internet Research, 11(4).
Chun, S. A., Shulman, S., Sandoval, R., & Hovy, E. (2010). Government 2.0: Making
connections between citizens, data and government. Information Polity, 15(1), 1-
54
9.
Cybenko, G., & Brewington, B. (1999). The foundations of information push and pull. In
The mathematics of information coding, extraction and distribution (pp. 9-30).
New York, NJ: Springer.
DeLone, W. H., & McLean, E. R. (1992). Information systems success: The quest for the
dependent variable. Information Systems Research, 3(1), 60-95.
Duggan, M., Ellison, N. B., Lampe, C., Lenhart, A., & Madden, M. (2015). Social media
update 2014. Retrieved from
http://www.pewinternet.org/2015/01/09/socialmedia-update-2014/
Dutta-Bergman, M. J. (2004). Primary sources of health information: Comparisons in the
domain of health attitudes, health cognitions, and health behaviors. Health
Communication, 16(3), 273-288. Retrieved from
http://dx.doi.org/10.1207/S15327027HC1603_1
Fischhoff, B. (2012). Communicating risks and benefits: An evidence based user's guide.
Silver Spring, MD: FDA Department of Health and Human Services.
Flynn, D. (2014). USDA: U.S. foodborne illnesses cost more than $15.6 billion annually.
Retrieved from http://www.foodsafetynews.com/2014/10/foodborne-illnessescost-
usa-15-6-billion-annually/ - .VYE-IhNViko
Fox, C. J. (1983). Information and misinformation: An investigation of the notions of
information, misinformation, informing, and misinforming. Westport, CT:
55
Greenwood Press.
Frighetto, J., & Wolf, E. (2014). Global consumers more likely to buy new products from
familiar brands. Retrieved from
http://www.nielsen.com/us/en/pressroom/2013/global-consumers-more-likely-to-
buy-new-products-from-familiarb0.html
George Ettel Iii, I. N., Ettel, D., Wilson, C., & Meola, P. (2012). How do adolescents
access health information? And do they ask their physicians? The Permanente
Journal, 16(1), 35.
Glasgow, R. E., Klesges, L. M., Dzewaltowski, D. A., Estabrooks, P. A., & Vogt, T. M.
(2006). Evaluating the impact of health promotion programs: Using the RE-AIM
framework to form summary measures for decision making involving complex
issues. Health Education Research, 21(5), 688-694.
Hanafi, S., Kasim, M., Ibrahim, M., & Hancock, D. (2009). Business reporting on the
Internet: Development of a disclosure quality index. International Journal of
Business and Economics, 8(1), 55-79.
Harris, J. K., Mueller, N. L., & Snider, D. (2013). Social media adoption in local health
departments nationwide. American Journal of Public Health, 103(9), 1700-1707.
Huang, Z., & Benyoucef, M. (2013). From e-commerce to social commerce: A close look
at design features. Electronic Commerce Research and Applications, 12(4),
246259.
56
Java, A., Song, X., Finin, T., & Tseng, B. (2007). Why we twitter: Understanding
microblogging usage and communities. Paper presented at the 9th WebKDD and
1st SNA-KDD 2007 workshop on Web mining and social network analysis, San
Jose, CA.
Jin, G. Z., & Leslie, P. (2002). The effect of information on product quality: Evidence
from restaurant hygiene grade cards. Retrieved from
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=322883
Krikelas, J. (1983). Information-seeking behavior: Patterns and concepts. Drexel Library
Quarterly, 19(2), 5-20.
Kuhlthau, C. C. (1991). Inside the search process: Information seeking from the user's
perspective. Journal of the American Society for Information Science, 42(5),
361371.
Ma, J., Almanza, B., & Ghiselli, R. (2015). Social media’s potential as a communication
tool for health departments interested in providing food safety information to the
public. Paper presented at the 2015 Annual ICHRIE Summer Conference,
Orlando, FL.
Mou, Y., & Lin, C. A. (2014). Communicating food safety via the social media: The role
of knowledge and emotions on risk perception and prevention. Science
Communication, 36(5), 593-616.
O'Keeffe, G. S., & Clarke-Pearson, K. (2011). The impact of social media on children,
adolescents, and families. Pediatrics, 127(4), 800-804.
57
Park, C. W., & Lessig, V. P. (1981). Familiarity and its impact on consumer decision
biases and heuristics. Journal of Consumer Research, 8(2), 223-231.
Pew Research Center. (2015a). Health Fact Sheet. Retrieved from
http://www.pewinternet.org/fact-sheets/health-fact-sheet/
Pew Research Center. (2015b). Social Media Use by Age Group Over Time. Retrieved
from http://www.pewinternet.org/data-trend/social-media/social-media-use-byage-
group/
Pew Research Center. (2015c). Social Media User Demographics. Retrieved from
http://www.pewinternet.org/data-trend/social-media/social-media-
userdemographics/
Porucznik, C. A., & Royal DeLegge, M. P. A. (2013). Impact of internet posting of
restaurant inspection scores on critical violations. Journal of Environmental
Health, 75(10), 8-12.
Ramanadhan, S., Mendez, S., Rao, M., & Viswanath, K. (2013). Social media use by
community-based organizations conducting health promotion: A content analysis.
BMC Public Health, 13(1), 1129.
Seo, S., Jang, S. S., Almanza, B., Miao, L., & Behnke, C. (2014). The negative spillover
effect of food crises on restaurant firms: Did Jack in the Box really recover from
an E. coli scare? International Journal of Hospitality Management, 39, 107-121.
Shannon, C. E. (1949). The mathematical theory of communication. Urbana, IL:
University of Illinois Press.
58
Shrigley, R. L. (1978). The persuasive communication model: A theoretical approach for
attitude change in science education. Journal of Research in Science Teaching,
15(5), 335-341.
Snyder, L. B. (2001). How effective are mediated health campaigns. Public
Communication Campaigns, 3, 181-190.
Thackeray, R., Neiger, B., Smith, A., & Van Wagenen, S. (2012). Adoption and use of
social media among public health departments. BMC Public Health, 12(1), 242.
Tinker, T., & Fouse, D. (2009). Special report: Expert round table on social media and
risk communication during times of crisis: Strategic challenges and opportunities.
Trends in Food Science & Technology, 30(1), 84–91.
Trouten, D. J. (2013). From print to pixels: Exploitation of the Internet’s potential by
evangelical print periodical publishers. Retrieved from ProQuest Digital
Dissertations database.
Tversky, A., & Kahneman, D. (1973). Availability: A heuristic for judging frequency and
probability. Cognitive Psychology, 5(2), 207-232.
Vorvoreanu, M. (2004). Building and maintaining relationships online: A framework for
analyzing the public relations website experience. Retrieved from ProQuest
Digital Dissertations database.
Vorvoreanu, M. (2006). Online organization–public relationships: An experiencecentered
approach. Public Relations Review, 32(4), 395-401.
59
Vorvoreanu, M. (2008). Web site public relations: How corporations build and maintain
relationships online. Cambridge, NY: Cambria Press.
Wantland, D. J., Portillo, C. J., Holzemer, W. L., Slaughter, R., & McGhee, E. M. (2004).
The effectiveness of web-based vs. non-web-based interventions: A meta-analysis
of behavioral change outcomes. Journal of Medical Internet Research, 6(4).
Webb, T., Joseph, J., Yardley, L., & Michie, S. (2010). Using the internet to promote
health behavior change: a systematic review and meta-analysis of the impact of
theoretical basis, use of behavior change techniques, and mode of delivery on
efficacy. Journal of Medical Internet Research, 12(1), e4.
Wilson, T. D. (2005). Evolution in Information Behavior Modeling Wilson's Model. In K.
E. Fisher, S. Erdelez, & L. McKechnie (Eds.), Theories of information behavior
(pp. 31-38). Medford, N.J.: American Society for Information Science and
Technology.
Wixom, B. H., & Todd, P. A. (2005). A theoretical integration of user satisfaction and
technology acceptance. Information Systems Research, 16(1), 85-102.
Worsfold, D. (2006). Eating out: Consumer perceptions of food safety. International
Journal of Environmental Health Research, 16(3), 219-229.
Young, N. J., & Von Seggern, M. (2001). General information seeking in changing times:
A focus group study. Reference & User Services Quarterly, 41(2), 159-169.
60
CHAPTER 4. STUDY 2: HOW CONSUMERS INTERACT WITH WEBSITES TO
OBTAIN FOOD SAFETY INFORMATION: AN APPLICATION OF WEBSITE
EXPERIENCE ANALYSIS (WEA)
The following study will be submitted to a peer-reviewed journal and is
included in this non-traditional thesis/dissertation as chapter 4. This article is
written in the APA style format.
4.1 Abstract
Food safety communication is essential in protecting public health. But the role of the
media has been largely overlooked in food safety communication studies. This study
utilized Website Experience Analysis (WEA) and examined how website characteristics
are linked with users’ perceptions. Oregon’s health department website was examined by
participants who then answered questions about their website experience addressing key
efficacy perceptions. The data indicate that several website features are related to both
information and media characteristics, and were experienced by users as important and
influential in their perception and later behavioral intention formation. Additionally, this
study illustrates the utility of WEA for online food safety communication research.
Keywords: food safety, Internet communication, food safety communication, WEA,
efficacy, website features
61
4.2 Introduction
Websites are excellent platforms for communicating food safety information, particularly
about foodborne illness outbreaks. Websites have unique characteristics, such as
interactivity, speed, and relatively low cost; these make them very attractive as a media
food safety communication. With 59% of the United States adult population going online
for health information (Pew Research Center, 2015a), the use of websites to communicate
food safety information seems appropriate. Unfortunately, websites’ potential for food
safety communication has not been realized (Choi & Almanza, 2012; Ma et al., 2015).
Media’s role has been largely overlooked in food safety communication. Research has
focused on adoption scope and content posted (Bortree & Seltzer, 2009; Chapman et al.,
2014; Harris et al., 2013; Thackeray et al., 2012). Better utilization of websites to
communicate food safety information can only be accomplished if user-website
interactions are better understood. The purpose of this research is to utilize Website
Experience Analysis (WEA) to examine user experiences and link website characteristics
with users’ perceptions to improve understanding of food safety communication.
4.3 Literature Review
4.3.1 Food safety communication
Foodborne illness outbreaks can be devastating to the economy, society, and businesses.
The Centers for Disease Control and Prevention (CDC) estimated that one in six
Americans (48 million) gets sick annually, 128,000 are hospitalized and 3,000 lose their
lives because of foodborne diseases (CDC, 2014a). Food Safety News suggests that
foodborne illnesses cost the economy more than 15.6 billion dollars each year (Flynn,
62
2014). Even one foodborne illness outbreak can damage a company’s reputation so that it
will never fully recover (Seo et al., 2014). Almost three decades after their own E. coli
outbreak, the stock of Jack in the Box still suffers a price drop whenever an E.Coli
outbreak happens (Seo et al., 2014).
Communication of food safety information can be powerful in combatting foodborne
illness outbreaks as communication can not only change consumer behavior (Choi et al.,
2011; Porucznik & Royal DeLegge, 2013), but also improve food safety control
(Almanza et al., 2002; Jin & Leslie, 2002). Traditionally, food safety communication has
had limited reach and impact (Charles & Lawrence, 1990; Dutta-Bergman, 2004). Newer
communication platforms, especially Internet-based platforms, present new possibilities
in food safety communication. Few studies have explored these possibilities. Studies
have shown that Internet platforms such as social media are starting to gain popularity
among health departments (Avery et al., 2010; Harris et al., 2013; Thackeray et al., 2012)
and other organizations such as the CDC, the US Food and Drug Administration (FDA),
the World Health Organization (WHO), and the American Public Health Association
(APHA) to communicate public health information (Heldman, Schindelar, & Weaver,
2013). But these studies primarily focused on examining social media’s adoption and
usage based on contents posted online (using traditional methods such as content
analysis).
4.3.2 Information seeking
In the literature of information-seeking behaviors, the models of Wilson (1981; 1999),
Ellis (1989, 2005), Krikelas (1983), Kuhlthau (1991), and Marchionini (1995) are
63
probably the most cited. Some models such as those by Ellis were tested to be relevant
and robust in electronic contexts, particularly in user Internet behavior studies
(Marchionini, 2008; 2003; van Deursen & van Dijk, 2009).
Although these models take different approaches, some dimensions consistently emerge
as critical in understanding the ways users seek information. All models consistently
include dimensions related to interface, including the rules and mechanisms for accessing
the information, as well as the quality and quantity of content. This highlights the
importance of both media and content in the online information-seeking experience.
Food safety communication has been moving online. The CDC, during the 2009
Salmonella typhimurium outbreak associated with peanut-containing products, effectively
engaged social media platforms to disseminate information and manage public outrage
and panic (CDC, 2012a).
In food safety and health communication, the characteristic of trustworthiness has a huge
impact on communication outcomes (Eysenbach, Powell, Kuss, & Sa, 2002).
Additionally, increasing Internet usage has amplified the need for high-quality
information (Moorhead et al., 2013; Rutsaert et al., 2013). The Internet has also changed
how people obtain general information about food. For example, Internet sites such as
Yelp and Tripadvisor have influenced where many people want to go for a meal (Gregory
& Kim, 2004; Parikh, 2013; Resnick, Zeckhauser, Friedman, & Kuwabara, 2000).
Although the Internet holds great potential for food safety communication, there is a gap
in the research on how to best provide food safety information. Thus, this study focuses
on both information and media characteristics in order to better understand food safety
64
communication. Improvements in food safety communication require a good
understanding of the mechanisms influencing individuals’ information-seeking behaviors
and how website features, such as the information or media characteristics, link to uses
perceptions.
4.3.3 Website Experience Analysis
Website Experience Analysis (WEA) was created to explore users’ experience and
interpretation of websites. WEA shifts the focus from the content posted on a website to a
user’s experience using the website. The reasons for adopting a user-centered approach
are grounded in communication theories. Meaning is not embedded in the text, waiting to
be discovered by users. Meaning is created in the process of reading and interacting with
the website (Fish, 1980). Thus, analyses based on texts (such as content analysis) should
be replaced by observation and analysis of people’s experience and perception/attitude
changes while using and interacting with the platforms (Vorvoreanu, 2004).
WEA requires participants to examine a website and answer questions along the way
about their experience (Vorvoreanu, 2008b). The format of the questions was derived
from Prominence-interpretation Theory (Fogg, 2003). Prominence-interpretation theory
states that evaluation and usage of a website, credibility assessments in particular, are
influenced by the prominence of website elements (if the elements are noticed) and the
interpretation of such elements by users. WEA allows questions to be asked that will not
direct the participant’s attention to certain website elements, but rather focuses questions
on user’s perceptions; this way, both preeminence and interpretation can be assessed. And
65
this is perhaps the biggest advantage of WEA – the ability to map and link website
elements to users’ perceptions.
In this study, four perceptions were assessed: self-efficacy, response efficacy, process
efficacy, and information efficacy. In applying WEA, a pair of questions is used for each
type of efficacy. The first question asks participants to provide a numeric rating for their
perception about that efficacy. The second question asks for website elements that the
perception is based upon. For example, the pair of questions used for self-efficacy was:
“Do you feel you are able to locate the needed information?” (on a scale of 1-10, with 1 =
not at all and 10 = very much); and “What on the website makes you feel this way?”
(open-ended question).
4.3.4 Efficacy perceptions
Efficacy is a core perception that is predictive of communication outcomes including
attitudes and behaviors. Self-efficacy can be defined as one's belief in one's ability to
accomplish tasks and achieve a desired outcome (Bandura, 1977, 2010; Witte & Allen,
2000). The concept of self-efficacy has appeared widely in communication literature such
as the Health Belief Model (Schiavo, 2014), Protection Motivation Theory (Maddux &
Rogers, 1983; Rogers, 1975), and the Extended Parallel Process Model (EPPM) (Witte,
1992). Self-efficacy was also empirically tested and found to be influential in determining
communication outcomes (Maloney et al., 2011; Webb et al., 2010; Witte & Allen, 2000).
Another important concept in predicting communication outcomes is response efficacy.
Response efficacy represents one’s belief that a certain action will be effective. While
self-efficacy is concerned with how competent we feel, response efficacy is whether we
think our actions can lead to the desired results (Floyd, Prentice‐
66
dunn, & Rogers, 2000). Response efficacy also has been studied extensively, and its
importance is well established (Witte & Allen, 2000).
Following previous literature, self- and response efficacy were adopted in understanding
consumer interaction with websites. In the context of this study (food safety
communication), self-efficacy and response efficacy were defined as one’s belief in his or
her ability to find the needed information and the extent to which one believes that
looking up information will help to reduce the risk of getting sick. Here response efficacy
is assessed after message exposure, which is slightly different from its traditional
applications. In its original form, response efficacy is widely used in understanding
message components, which is why it is defined as one’s belief that a recommendation
action (e.g. washing hands – usually embedded in the message) will reduce risks (e.g.
prevent illness). But along with development and advancement of EPPM, response
efficacy has been extended and applied in a wide range of contexts such as assessing
attitudes towards an action (Barnett et al., 2009) and predicting behaviors (Hullett &
Witte, 2001; Roberto & Goodall, 2009). Thus, this study assesses one’s belief in whether
looking up information will help to reduce the risk of getting sick.
It is without doubt that Internet communication tools have changed individual’s
information-seeking process, but such impacts can be mixed. On one hand, the Internet
allows the information to be more accessible, reducing the efforts required to obtain the
information; but on the other hand, the quality of information has increasingly become a
concern as anyone can share information online. As noted by Marchionini (2008), digital
technology “changes the very nature of information and how people interact with each
other and with information.” Thus, it only seems reasonable to look at interactions among
67
information, people, and technologies. In this sense, it is important to examine how the
media (e.g. website) and the information intermingle and influence consumers’
information-seeking behaviors. When it comes to media selection, the Principle of Least
Effort (PLE) suggests that information seekers prefer to use media that requires the least
amount of effort or is easiest to use (Case, 2005). In understanding information’s
influence, Diffusion of Innovation Theory (diffusion theory), although initially designed
to examine adoption of technological innovations, can be extended to study
informationseeking behaviors (Lajoie-Paquette, 2005; Rogers, 2003). In essence,
diffusion theory highlights the phenomena that perceived relative advantages and
disadvantages or the perception of value and usefulness of information is the determinant
in individual’s information seeking behaviors. But these perceptions towards media and
information cannot be fully captured by self- and response efficacy. Thus, to
conceptualize people’s perceptions towards media and information beyond what can be
accounted for by self- and response efficacy, this study uses two concepts - “process
efficacy” and “information efficacy” to represent the impact of media and information in
an individual’s information-seeking experience. More specifically, process efficacy is
defined as one’s belief that the process can be easily done, and information efficacy is
defined as one’s belief that the information will be helpful. In sum, this research focuses
on four dimensions of efficacy perception: self-efficacy, response efficacy, process
efficacy and information efficacy in studying people’s information seeking behavior.
With these in mind, this study attempts to address two research questions: 1. How do
consumers get foodborne illness outbreak information? 2. What website characteristics
68
are experienced by consumers that influence their efficacy perceptions (response, self,
process, and information) towards the communication?
4.4 Methodology
4.4.1 Website selection
WEA is “very well suited for examining a single website with the purpose of
understanding and improving the experience of using it” (Vorvoreanu, 2008b). A health
department website was selected for this study because health departments are perfectly
positioned to communicate information about foodborne illness outbreak (e.g. have more
localized information than CDC), and in fact, have the primary responsibility for
identifying and investigating foodborne illness outbreaks (Lynch, Painter, Woodruff, &
Braden, 2006). After examining the websites of numerous health departments, Oregon
was chosen because it is comparatively competent in using website for foodborne illness
outbreak communications. The State of Oregon health department is one of six Integrated
Food Safety Centers of Excellence that were established through a competitive process to
improve foodborne illness surveillance after the Food Safety Modernization Act (FSMA)
came into effect. The six selected state health departments can be considered highly
competent in foodborne illness surveillance and investigations. Among the six centers -
Colorado, Florida, Minnesota, New York, Oregon, and Tennessee (CDC, 2015b), Oregon
was chosen because it is the only one that currently posts updates on ongoing foodborne
illness outbreaks (the latest post was within the last three months). In addition, lessons
learned from this website analysis would be expected to provide an exemplar for other
health departments as they are based on an actual model of a health department website.
69
4.4.2 Sample selection
In using WEA, it is highly desirable for the samples to be homogenous – that is, the
participants are members of the same public/interpretive community (Fish, 1980; Machor
& Goldstein, 2001; Vorvoreanu, 2006). While WEA can be conducted with different
publics, it is important to know that the experience of each public can be different using
the same website. Research has suggested that age may influence individuals’ usage of
the Internet and their perception of the usefulness of Internet-based services (Jones,
Cassie, Thompson, Atherton, & Leslie, 2014; van Deursen & van Dijk, 2014; Zickuhr &
Madden, 2012). The targeted populations, in terms of age, are younger adults (18-29) and
young adults (29-40). The reason for this decision is the Internet leverages the most value
when communicating food safety information with younger and young adults. These two
age groups are historically hard-to-reach population groups in health communication
(Gray, Klein, Noyce, Sesselberg, & Cantrill, 2005), but because they frequently use the
Internet for information (Pew Research Center, 2010, 2015b), using Internet platforms for
food safety communication has great potential to reach these populations.
Additionally, the level of education may influence individuals’ digital skills and thus
influence how they interact with websites (Katz & Aspden, 1997; van Deursen & van
Dijk, 2014). With this in mind, a quota sampling strategy was applied using age and
education level as strata. For each of the subgroups (e.g. age 18-25, with a bachelor’s
degree or more), four participants (Vorvoreanu, 2008b) were recruited. See Table 4-1.
Table 4-1. Sample distribution based on age and education
Age
Education
70
18-28 Bachelor’s and above (4) Less than Bachelor’s (4) 29-40 Bachelor’s and above
(4) Less than Bachelor’s (4)
To collect data that includes a range of ages and education levels, two different settings, a
public library and a university campus, were utilized. A pre-participation survey was then
utilized to capture potential participants’ ages and education levels. In total, 16
participants were recruited (see Table 4-1 for the distribution).
4.4.3 Procedures
During each one-on-one research session (around 40 to 60 minutes), participants were
asked to describe their typical behaviors in getting foodborne illness outbreak
information. Participants were then asked to look for information about a recent
foodborne illness outbreak using the State of Oregon’s health department’s website and
answer pairs of questions addressing the four efficacy perceptions (response, self,
process, and information). In each pair, the first question asked a numeric rating for the
efficacy perception while the second, open-ended question asked participants to explain
what website aspects accounted for their opinion and perception. To ensure all
participants’ experiences were identical, the cookies and temporary Internet files were all
cleaned after each data collection session. Recurring themes were identified in the
responses to each open-ended question, and the answers were thematically coded for the
presence of these themes.
4.5 Results
Of the 16 participants whose responses were recorded and analyzed, 8 were female and 8
were male. Overall, all participants had quite a lot of experience using both computers
71
(M= 14.375 years) and the Internet (M= 14.125 years). Participants reported having used
computers for 8-30 years, and the Internet for 8-30 years. All participants reported using
computers and the Internet frequently, at least once a day.
4.5.1 Two-stage process in seeking out foodborne illness outbreak information
When asked to describe their typical information behavior regarding foodborne illness
outbreak information, 14 participants indicated that they do not look for foodborne illness
outbreak information regularly. The other two participants, who both happen to fall in the
group age between 29-40 with a bachelor’s degree or above, indicated that they sign up
for newsletters from different agencies, including CDC, Food Safety News, and FDA, for
updates about foodborne illness outbreaks. Participants’ information needs to get regular
foodborne illness outbreak updates corresponded to their interest and concern for food
safety. Individuals who were more interested in and concerned about food safety received
updates on outbreaks regularly; while the rest of the participants were more passive
consumers of information. They received foodborne illness outbreak information along
with other news, and they did not have a particular interest in getting regular updates.
One participant made a comment to illustrate this point: “If there’s something happening
that is serious, I will see it in the news. If it is not on the news, I think it is not to the point
for me to be concerned.”
On the other hand, all participants indicated that once they became aware of an instance
or outbreak, they would look for more information if the event was personally relevant
(e.g. geographically relevant and ongoing). “If it is happening in my area, like in the
restaurant I know of, I will definitely look for more information.” This suggests that a
72
risk is only perceived if there’s an ongoing and relevant (personal risk) outbreak and only
when a risk is perceived is the need for information generated.
Individuals have different ways of becoming aware of an outbreak. The people who were
more concerned had signed up to receive regular updates (e.g. newsletters), but the
majority of respondents passively waited for the information to be pushed to them (e.g.
through TV, newspapers, or social media updates). To learn about if there’s an outbreak, a
number (8) of participants indicated that social media was their most preferred way of
getting updates, followed by phone applications (4), TV (3), and radio (1).
After learning that there’s an outbreak, the uncertainties and potential risk presented
create a drive or need for information. And this drive motivates individuals to seek out
more information to reduce the uncertainties and protect themselves. In seeking more
information, all participants indicated that they would start with a search engine (e.g.
Google, MSN, or Yahoo) and not go to a specific website for information. The reasons for
this are that participants feel search engines provide more up-to-date information that is
directly related to the topic of interest. In addition, search engines provided a list of sites
individuals can select from. So with the least amount of effort, search engines provided a
“comprehensive” list, within which at least one site would likely be able to supply useful
information. Additionally, younger participants (both groups ages 18-30) were more
likely to use smart phones instead of computers when they wanted to look up
additional information.
4.5.2 Mapping perceptions to website characteristics
73
Self-efficacy
The first pair of questions, “Do you feel you are able to locate the needed information?”
and “What on the website makes you feel this way?” were used to address self-efficacy
perception. On average, the participants gave self-efficacy a score of 6.7 (out of 10). The
main website feature that participants indicated as important for them in formation of the
self-efficacy perception was information organization, including the use of site a map,
tabs, and structural highlights. The participant’s comments (along with a screenshot,
Figure 4-1) illustrates the importance of an information organization structure:
The website is very organized. It has multiple colored tabs with different topics…
this structure, with the row representing topics and the column representing
services. I feel that I can find the information I need here.
Figure 4-1. Screenshot of Oregon state health department website illustrating participant
comments on information organization
74
Another important website feature that emerged from participants’ comments was the
links on the website. This is somewhat related to the organization of the website, yet it is
specifically mentioned by multiple participants (see figure 4-2).
I feel these links help me to find the needed information.
Figure 4-2. Screenshot of Oregon state health department website illustrating participant
comments on the use of links
One other important website aspect related to self-efficacy was specificity of information.
When information on the websites was very specific, participants were more confident
that they could find the information needed.
I am disappointed that a lot of the information is just too general. I am not sure if I
will be able to find what I need.
The last important theme to emerge was that self-efficacy was influenced and possibly
more heavily by pre-existing beliefs that would not be changed directly by interacting
with the website. These pre-existing beliefs included one’s confidence in his or her ability
75
to use Internet-based applications and one’s existing feeling towards food safety
communication platforms.
Even if you give a very crappy website, I can still find what I need… I use the
Internet every day for my work.
Even though this website looks really good, I know websites like this normally do
not work as well as they appear to be. So, I don’t know if I can find the needed
information.
Response efficacy
The next pair of questions asked participants the extent to which they believe that looking
up information on this website will help to reduce the risk of getting sick and what
website characteristics made them feel that way. The specific questions were: “Do you
feel that looking up food safety information will help you reduce the risk of getting sick?”
and “What on the website makes you feel this way?” Overall, all participants give an
average score of 7.0 (out of 10) for response efficacy.
One important theme related to response efficacy was the breadth of information
provided. The participants indicated that using a website that provides a wide range of
information will make them trust the information more and thus feel a stronger
connection between looking up information and staying healthy.
This website provides a lot of information covering a number of different topics. I
feel that if they can put up so much information, they must be the experts in this.
76
And because I will trust them more, I feel that looking up information on this
website will help me stay healthy.
Another theme that emerged related to response efficacy was the visual aspects of the
website. Instead of an element, this feature is users’ interpretation of visual presentations
on the website or visual esthetics. Attractive graphics, coordinated colors, and visually
appealing designs can all be considered as important aspects of visual esthetics.
The picture looks very happy and makes me think that they care about our
welfare. All the different colors they use are coordinated and not too
overwhelming, so I feel that they put in efforts in making this website look nice…
I trust the information more.
It is interesting to note that most website characteristic emerged here are related to trust in
the website. Belief that looking up information will help to reduce the risk of getting sick
may in fact be a trust issue. Further investigations are needed before any affirmative
conclusions can be reached.
The final major theme related to response efficacy was that although website
characteristics were important, experience with the website alone would not determine
users’ perception of response efficacy. Response efficacy is also influenced by factors
such as existing knowledge, previous experience with information, and confidence in
one’s health that are outside the interaction with the website.
They lay out things very nicely and clearly for me to see… But I don’t know
much about this E.coli thing… I am not sure if this information will protect me.
77
I really don’t think anything on this website changes my opinion. I would totally
still look for information if I know there’s an outbreak.
Process efficacy
The third pair of questions was intended to gauge users’ process efficacy perception and
the reasons behind their perceptions. Two questions were used: “Do you feel that the
process of locating needed information is easy?” and “What on the website makes you
feel this way?”. The average score for process efficacy from all participants is 5.1 (out of
10).
The presence of a search box is the most important feature related to process efficacy, as
100% of participants touched upon this aspect.
Having a search box makes it so much easier for me to find the information I
need.
However, 14 out of 16 participants mentioned that the search experience was not
satisfactory, which is probably why the average score for process efficacy is
comparatively lower. The major problems included: the search results cannot be
organized according to different criteria (such as time posted), the search function does
not fully support the use of keywords, redundant layout (e.g. multiple search boxes close
to each other, see Figure 4-1), and inability to refine searches.
I cannot sort accordingly to time.
When I typed “food borne illness outbreak” in the search box, no results were
found.
78
There are two search boxes? … I will just go ahead and use one and see what
happens.
The first five links are all about inspections, educational, and training materials...I
cannot filter out the non-relevant stuff and only focus on news updates of
outbreaks.
The second website feature experienced by participants as important to process efficacy
was interactivity. And more specifically, interactions with other humans such as different
users and the source of information were indicated to be important.
Being able to share this information right from this page makes it so easy to use
this information.
I love that there is a comment box… So if I cannot find what I need right away, I
can always ask and find out later.
The third feature on a website that was mentioned frequently by participants was the
presence of links. The links that direct users to additional information – such as
explanations of a concept and references from outside sources – were perceived to make
it easy for users to locate needed information.
I can click through these links and easily find a lot of information.
The fourth website aspect mentioned by participants was the use of visuals. Visuals are
pictures and videos that facilitate users’ understanding of the information presented.
Watching this video really makes it easy for me to understand what was going on.
79
The pictures really helped me in seeing where the outbreak is spreading.
However, it is important to note that using pictures and videos from a third party site can
be risky as security is not guaranteed. To maintain the image that the website is a safe and
credible source for food safety information, the pictures and videos need to be carefully
curated.
I very much enjoyed watching this video… But after the video is done playing, it
links to some other videos that are apparently not ‘legitimate.’
Information efficacy
The last pair of questions asked about participants’ information efficacy perception and
the website characteristics that influence that perception. “Do you feel that information
communicated is helpful to you in making dining choices?” and “What on the website
makes you feel this way?” were used to assessing information efficacy. The average score
for information efficacy was calculated to be 6.9 (out of 10).
The most frequently mentioned theme was timeliness of the information. All participants
think current and up-to-date is essential for the information about the foodborne illness
outbreak to be useful.
I see that the updates are current… so I feel they can be useful.
The second website aspect that emerged was source. The source directly related to
consumer perception of trustworthiness. All participants noticed that the website was a
80
government website and because of this, they perceived it to be trustworthy and credible
– and thus considered the information on the website to be more useful.
I trust it … the information is coming from a legitimate government agency.
The third important theme was information accuracy. To participants, if the information
was specific and contained reasonable recommendations, it was perceived to be accurate
and thus useful.
The information tells me exactly which brand is involved and which food item is
contaminated. I like this. I can totally use this information.
The fourth theme to come out of participants’ comments was user orientation, which was
reflected in certain interactive features of the websites. User orientation is similar to what
in marketing would be defined as customer interest. Participants felt that paying attention
to users’ needs indicates that a website (and the entity using it) cares about users, and thus
the information communicated can potentially be more useful. Two main website
features related to individual perception of user orientation were the ability to share
information with other users and the capacity to provide feedback, ask questions, and get
responses. In other words, interactivity – interaction with other users and the information
source – is highly important for consumers. This website aspect was also mentioned
frequently when process efficacy was evaluated.
Mentioned less frequently, but still appearing multiple times, was the website’s
readability. Here the results were conflicting, while some participants did not appreciate
technical terms (at least not without explanations), others preferred the presence of
81
technical terms. But overall, participants all welcomed information that was clearly
written and easily understandable.
These professional terms make me feel that they know what they are doing.
I don’t have a lot of background in this area… I don’t like when they use words I
cannot understand.
I like this is clearly written with short and concise sentences… I think it will be
helpful.
Other website aspects mentioned by participants, such as no advertisements, were not
mentioned frequently enough to be considered major themes. One interesting finding, as
mentioned in two participants’ comments, was that the quantity of information presented
would influence their perceived usefulness of information. However, one participant
noted the more information, the better while the other indicated the opposite.
A summary of the linkages between website aspects and user efficacy perceptions were
presented in Table 4-2.
Table
2
-2. Participant experience map - a summary of efficacy scores and website aspects linked
to different efficacy perceptions
2
.5.3 Design suggestions
Some design suggestions were also mentioned by participants. First, the majority (14 out
of 16) of the participants mentioned that they would prefer an “alert feature” that allows
82
Efficacy Score(Out of 10) Website aspects linked Other aspect linked
Self 6.7 Information organization
Links
Specificity of information
Pre-existing beliefs
Response
7
Breadth of information
Visual esthetics
Knowledge,
experience, and
confidence
Process
5.1
Search box
Interactivity
Links
Visuals (pictures and videos)
Information
6.9
Information timeliness
Source
Information accuracy
User orientation (the ability to share
information with other users and the
capacity to provide feedback, ask
questions, and get responses)
Readability
concise is a good practice, but additional features, such as an explanation box, can be
used when users hover their mouse over certain difficult-to-understand concepts. Such
feature can also be used to explain how some website functions work. Third, making the
website mobile friendly seems to be an important consideration as a large number of
quick grasp of the most recent updates. For example, a pop-up alert on the home page, an
easy-to-spot link on the first page (with a different color or otherwise emphasized), or a
floating window (which remains in sight when users scroll up and down) would all be
possible ways to communicate the most recent updates. Second, to make information
83
consumers used their phones for information. Fourth, adding a function for users to
enlarge font sizes may be helpful.
4.6 Discussion And Impact
When discussing the implications, it is important to keep in mind the limitations of this
study. First, this study utilized younger populations, and there’s no evidence that these
results could be generalized to all populations. Second, this study utilized one website in
data collection. Additional features that are also important may be absent from this
particular website and thus not mentioned by participants. In the future, a comparison
across different platforms, including both websites and social media could be done to
verify the robustness of the results.
4.6.1 Implications for food safety communication practice
The observed information seeking behaviors where individuals become aware of an
outbreak, interpret the risks and get motivated to seek information, and use their preferred
media to obtain needed information indicates a two-stage process in foodborne illness
outbreak information-seeking. The first stage is initial exposure and the second stage is
conscious and active information seeking. This means that when drafting communication
strategies, two questions needed to be considered: how do you reach broader audiences
when there’s an outbreak and how can the information seeking process be improved when
consumers are looking for additional information about an outbreak. Social media can be
helpful in reaching broader audiences as it can be used to post updates and push outbreak
information to consumers (alerting consumers). Social media allows entities to leverage
consumers’ existing network (e.g. a post can be shared on social media with friends and
84
family) as well as get in contact with younger, previously hard to reach, consumer groups
for food safety communication. This means that aside from being a potential vehicle to
build relationships with the public (Chapman et al., 2014), social media can be an
effective supplement to existing communication channels for foodborne illness outbreaks.
Additionally, traditional media (e.g. TV, newspaper, or radio) can also be leveraged at this
stage to reach more audience, particularly since the Internet - while covering the majority
of the population - is still not accessible by everyone.
As seen, after the initial exposure, consumers look for additional information about an
outbreak. During this process, consumers’ preferences and needs may shift and thus
require a different communication strategy to be implemented. Here, consumers are in
need of information to take action and protect themselves, information quality -
characterized by trustworthiness, specificity (details), and timeliness - becomes more
important. As such, website may become more suitable as social media starts to lose its
relative advantages at this stage of communication. To leverage the benefits of different
platforms, a strategy that integrates the usage of different platforms may in fact be more
appropriate. For example, using social media to post updates and alert consumers about
ongoing outbreaks. And use links to direct consumers to websites that contain more
detailed information. In the future, more exploration may be needed to suggest more
effective use of different platforms especially the combination of different platforms.
Additionally, it was discovered that, at the second stage, consumers commonly identify
the websites for information through the use of search engines. This highlights the need
for government and health agencies that are interested in communicating food safety
information to the public to do their best to get “found”. Overall, good practices include
85
adding more relevant keywords, providing timely updates, and incorporating links from
other trustworthy sources.
Furthermore, individuals only perceive a risk if a foodborne illness outbreak is personally
relevant to them. This implies that more localized and timely (e.g. real-time) updates may
be needed in communicating about foodborne illness outbreaks. As such, health
departments may be uniquely positioned to be the center for communications of
foodborne illness outbreaks as they are, in general, more aware of local events and have
more specific information for consumers.
It is possible that a user’s interpretation of a website is very different from the intended
meaning of website creators. WEA offers an easy way to find out about such potential
differences (Vorvoreanu, 2006). Utilizing WEA, three information characteristics –
trustworthiness, timeliness, and accuracy, along with three media characteristics –
searchability, interactivity, and enhanced usability (including visuals and links) were
found to be most important when users form their perceptions towards the website and
the information communicated. This has a number of design and website usage
implications. It is important to keep in mind the context when reading and using these
design implications. Although insightful, such design observations should be evaluated
within the context of this study as participants’ reactions were collected utilizing the
Oregon state health department’s website. Considering these observations as contextual
insights, they are not intended for generalization, but rather for understanding consumer
experience.
86
1. It is essential to help users reduce or minimize the efforts needed to look for
information. Having a working search function is probably the first thing entities may
want to consider. A search function is considered working if it allows users to quickly
locate the relevant information when they know little about the outbreak. So features
such as ‘fuzzy’ keyword search, filters allowing information to be ranked by the
number of clicks, time, and location, and advanced search supporting use of location
service may be valuable to incorporate.
2. Media and information characteristics interact and together create perceptions of
efficacy. For example, users derive meaning and make decisions not only based on
content elements, but also from the ease of accessing them. Thus, when designing
websites for food safety communication, consider carefully how the interaction of
content, layout, organization, and navigation can be leveraged. One example is to
make the most current updates appear at places that are easily seen, such as the top of
the page. Alternatively, entities can use a pop up banner/box to draw users’ attention
to the most current outbreak updates. Additionally, entities may want to consider
using a site map such as different colored tabs on the side or at the top of the website
to help direct users attention and support quick location of needed information.
3. Because users value interactivity, organizations such as health departments may want
to consider inviting the public to communicate with them and with each other. For
example, embed functions to share information on the site and provide means such as
a comment box for feedback. This way, websites can be used as a vehicle for building
relationships and sustaining engagements.
87
4. Visual elements are important in assisting users’ interpretation of information
communicated. When possible, incorporate relevant and visually appealing pictures
and graphs on the page and start to use videos. But make sure these pictures and
videos are carefully curated so that they do not link to unintended information as in
the case of most YouTube videos. Consumers’ trust, once broken, is hard to restore.
Hosting videos rather than embed from YouTube to avoid unintended
recommendations from YouTube, which can quite unpredictable, may be a wise thing
to do.
5. Some meta-communication of website elements should be kept in mind when
designing and using websites. Individuals derive meaning and make inferences about
the information, the website, and even organization from subtle aspects of website
design, such as the colors, the placement/ location of information, and pictures used.
Ideally, when setting up a website, a master organization plan, which includes topic
organization, color theme, placements of sections etc. need to be thoroughly thought
out. Again, it is important to help reduce users’ efforts in information seeking. For
example, when possible, put the most recent outbreak in an area that is more
noticeable (e.g. on the first page, and at the top of the page) and use color to highlight
it. It is advised to avoid assuming that if the information is there, users will find it.
Additionally, the conventional meanings of different design elements are good to
keep in mind. For example, the color red may signal ‘alert’ and ‘emergency’ while the
color green may signal ‘safe’ and ‘good’. Using colors to present the urgency levels
of different outbreaks may be helpful.
88
4.6.2 Implications for food safety communication research
This study is one of the first to demonstrate the importance of user-centered approaches
in studying food safety communication problems. Additionally, the results of study
suggest that the four efficacy perceptions may not work in a parallel fashion towards
communication outcomes. Furthermore, the formation of different types of efficacy
perceptions may not be distinct either, as some of the website features influenced more
than one of the efficacy perceptions. It is also interesting to note that information
characteristics may influence more than information efficacy, and media characteristics
may influence more perceptions than process efficacy. For example, interactivity – the
ability to share information with others and ask questions of the information source –
impacts both information and process efficacy. This further suggests that the relationship
between the four efficacy components can be very complex. Thus, it is important, in the
future, to further investigate the formation of these efficacy perceptions and examine the
relationships between them and how they interact in influencing communication
outcomes. Lastly, this study shed lights on the experience of the targeted interpretative
communities/ homogenous groups; future research can utilize the research procedures
outlined in this study and expand our understanding on the experience of different
interpretative communities (such as older users).
4.7 Conclusion
This study started with the fundamental question of how websites can be better utilized in
food safety communication. A user-centered research protocol, WEA was used to link
website characteristics with user perceptions. Data collected using the State of Oregon’s
89
health department website was presented. The results suggest that individuals go through
a two-stage process in obtaining foodborne illness outbreak information. Additionally,
this study found the website characteristics (trustworthiness, timeliness, accuracy,
searchability, interactivity, and enhanced usability) are associated with four kinds of
efficacy (self, response, process, and information). This study also illustrates the value of
WEA both in food safety communication practice and research. Lastly, as one of the first
attempts to understand the mechanisms behind food safety information behavior, this
study presented direction for future research, particularly focusing on recognizing and
refining the concepts of efficacy.
4.8 References
Almanza, B. A., Ismail, J., & Mills, J. E. (2002). The impact of publishing foodservice
inspection scores. Journal of Foodservice Business Research, 5(4), 45-62.
Avery, E., Lariscy, R., Amador, E., Ickowitz, T., Primm, C., & Taylor, A. (2010).
Diffusion of social media among public relations practitioners in health
departments across various community population sizes. Journal of Public
Relations Research, 22(3), 336-358. Retrieved from
http://www.tandfonline.com/doi/abs/10.1080/10627261003614427
Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavioral change.
Psychological Review, 84(2), 191-215.
Bandura, A. (2010). Self-Efficacy. Hoboken, NJ: John Wiley & Sons, Inc.
90
Barnett, D. J., Balicer, R. D., Thompson, C. B., Storey, J. D., Omer, S. B., Semon, N. L., .
. . Lanza, K. M. (2009). Assessment of local public health workers’ willingness to
respond to pandemic influenza through application of the extended parallel
process model. Plos One, 4(7), e6365.
Bortree, D. S., & Seltzer, T. (2009). Dialogic strategies and outcomes: An analysis of
environmental advocacy groups' facebook profiles. Public Relations Review,
35(3), 317-319.
Case, D. O. (2005). Principle of Least Effort In K. E. Fisher, Sansa Erdelez, & L.
McKechnie (Eds.), Theories of Information Behavior (pp. 289-292). Medford, NJ:
American Society for Information Science and Technology
CDC. (2011). 2011 Estimates of Foodborne Illness in the United States. Retrieved from
http://www.cdc.gov/foodborneburden/
CDC. (2012). The health communicator’s social media toolkit. Retrieved from
http://www.cdc.gov/healthcommunication/toolstemplates/socialmediatoolkit_bm.
pdf
CDC. (2015). Integrated food safety centers of excellence. Retrieved from
http://www.cdc.gov/foodsafety/centers/about.html
Chapman, B., Raymond, B., & Powell, D. (2014). Potential of social media as a tool to
combat foodborne illness. Perspectives in Public Health, 134(4), 225-230.
Charles, K. A., & Lawrence, W. (1990). Mass communication and public health:
Complexities and conflicts. Newbury Park, CA: Sage.
91
Choi, J., & Almanza, B. (2012). Health department websites as a source of restaurant
food safety information. Journal of Culinary Science & Technology, 10(1), 40-52.
Choi, J., Nelson, D. C., & Almanza, B. (2011). The impact of inspection reports on
consumer behavior: A pilot study. Food Control, 22(6), 862-868.
Dutta-Bergman, M. J. (2004). Primary sources of health information: Comparisons in the
domain of health attitudes, health cognitions, and health behaviors. Health
Communication, 16(3), 273-288.
Ellis, D. (1989). A behavioural approach to information retrieval system design. Journal
of Documentation, 45(3), 171-212.
Ellis, D. (2005). Ellis's model of information-seeking behavior. In K. E. Fisher, S.
Erdelez, & L. McKechnie (Eds.), Theories of information behavior (pp. 128-142).
Medford, NJ: American Society for Information Science and Technology
Eysenbach, G., Powell, J., Kuss, O., & Sa, E.-R. (2002). Empirical studies assessing the
quality of health information for consumers on the World Wide Web: A systematic
review. The Journal of the American Medical Association, 287(20), 2691-2700.
Fish, S. E. (1980). Is there a text in this class?: The authority of interpretive communities.
Cambridge, MA: Harvard University Press.
Floyd, D. L., Prentice‐ Dunn, S., & Rogers, R. W. (2000). A meta‐ analysis of research on
protection motivation theory. Journal of Applied Social Psychology, 30(2), 407-
429.
92
Flynn, D. (2014). USDA: U.S. foodborne illnesses cost more than $15.6 billion annually.
Retrieved from http://www.foodsafetynews.com/2014/10/foodborne-illnessescost-
usa-15-6-billion-annually/ - .VYE-IhNViko
Fogg, B. J. (2003). Prominence-interpretation theory: Explaining how people assess
credibility online. Paper presented at the CHI EA '03, Fort Lauderdale, FL.
Gray, N. J., Klein, J. D., Noyce, P. R., Sesselberg, T. S., & Cantrill, J. A. (2005). Health
information-seeking behaviour in adolescence: The place of the Internet. Social
Science & Medicine, 60(7), 1467-1478.
Gregory, S., & Kim, J. (2004). Restaurant choice. Journal of Foodservice Business
Research, 7(1), 81-95. Retrieved from http://dx.doi.org/10.1300/J369v07n01_06
Harris, J. K., Mueller, N. L., & Snider, D. (2013). Social media adoption in local health
departments nationwide. American Journal of Public Health, 103(9), 1700-1707.
Heldman, A. B., Schindelar, J., & Weaver, J. B., III. (2013). Social media engagement and
public health communication: Implications for public health organizations being
truly "social". Public Health Reviews, 35(1), 1-18.
Hullett, C. R., & Witte, K. (2001). Predicting intercultural adaptation and isolation: Using
the extended parallel process model to test anxiety/uncertainty management
theory. International Journal of Intercultural Relations, 25(2), 125-139.
93
Jin, G. Z., & Leslie, P. (2002). The effect of information on product quality: Evidence
from restaurant hygiene grade cards. Retrieved from
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=322883
Jones, J., Cassie, S., Thompson, M., Atherton, I., & Leslie, S. J. (2014). Delivering
healthcare information via the Internet: Cardiac patients' access, usage,
perceptions of usefulness, and website content preferences. TELEMEDICINE and
e-HEALTH, 20(3), 223-228.
Katz, J., & Aspden, P. (1997). Motivations for and barriers to Internet usage: Results of a
national public opinion survey. Internet Research, 7(3), 170-188.
Krikelas, J. (1983). Information-seeking behavior: Patterns and concepts. Drexel library
Quarterly, 19(2), 5-20.
Kuhlthau, C. C. (1991). Inside the search process: Information seeking from the user's
perspective. Journal of the American Society for Information Science, 42(5),
361371.
Lajoie-Paquette, D. (2005). Diffusion Theory In K. E. Fisher, Sansa Erdelez, & L.
McKechnie (Eds.), Theories of Information Behavior (pp. 118-122). Medford, NJ:
American Society for Information Science and Technology
Lynch, M., Painter, J., Woodruff, R., & Braden, C. (2006). Surveillance for foodborne
disease outbreaks: United States, 1998-2002. Atlanta, GA: US Department of
Health and Human Services.
94
Ma, J., Almanza, B., & Ghiselli, R. (2015). Social media’s potential as a communication
tool for health departments interested in providing food safety information to the
public. Paper presented at the 2015 Annual ICHRIE Summer Conference,
Orlando, FL.
Maddux, J. E., & Rogers, R. W. (1983). Protection motivation and self-efficacy: A revised
theory of fear appeals and attitude change. Journal of Experimental Social
Psychology, 19(5), 469-479.
Maloney, E. K., Lapinski, M. K., & Witte, K. (2011). Fear appeals and persuasion: A
review and update of the extended parallel process model. Social and Personality
Psychology Compass, 5(4), 206-219.
Marchionini, G. (1995). Information seeking in electronic environments. Cambridge, NY:
Cambridge University Press.
Marchionini, G. (2008). Human–information interaction research and development.
Library and Information Science Research, 30(3), 165-174.
Meho, L. I., & Tibbo, H. R. (2003). Modeling the information‐ seeking behavior of social
scientists: Ellis's study revisited. Journal of the American Society for Information
Science and Technology, 54(6), 570-587.
Moorhead, S. A., Hazlett, D. E., Harrison, L., Carroll, J. K., Irwin, A., & Hoving, C.
(2013). A new dimension of health care: systematic review of the uses, benefits,
95
and limitations of social media for health communication. Journal of Medical
Internet Research, 15(4).
Parikh, A. (2013). User generated restaurant reviews: Examining influence and
motivations for usage. Retrieved from ProQuest Digital Dissertations database.
Pew Research Center. (2010). Updated: Change in Internet access by age group,
20002010. Retrieved from http://www.pewinternet.org/2010/09/10/updated-
changein-internet-access-by-age-group-2000-2010/
Pew Research Center. (2015a). Health Fact Sheet. Retrieved from
http://www.pewinternet.org/fact-sheets/health-fact-sheet/
Pew Research Center. (2015b). Social Media Use by Age Group Over Time. Retrieved
from http://www.pewinternet.org/data-trend/social-media/social-media-use-byage-
group/
Porucznik, C. A., & Royal DeLegge, M. P. A. (2013). Impact of Internet posting of
restaurant inspection scores on critical violations. Journal of Environmental
Health, 75(10), 8-12.
Resnick, P., Zeckhauser, R., Friedman, E., & Kuwabara, K. (2000). Reputation systems.
Communications of the ACM, 43(12), 45-48. Retrieved from
http://search.ebscohost.com/login.aspx?direct=true&db=buh&AN=11887999&sit
e=ehost-live
96
Roberto, A. J., & Goodall, C. E. (2009). Using the extended parallel process model to
explain physicians' decisions to test their patients for kidney disease. Journal of
Health Communication, 14(4), 400-412.
Rogers, E. M. (2003). Diffusion of Innovations. New York, NY: Free Press.
Rogers, R. W. (1975). A protection motivation theory of fear appeals and attitude change.
The Journal of Psychology, 91(1), 93-114.
Rutsaert, P., Regan, Á., Pieniak, Z., McConnon, Á., Moss, A., Wall, P., & Verbeke, W.
(2013). The use of social media in food risk and benefit communication. Trends in
Food Science & Technology, 30(1), 84-91. Retrieved from
http://www.sciencedirect.com/science/article/pii/S0924224412002142
Schiavo, R. (2014). Health communication: From theory to practice. San Francisco, CA:
Jossey-Bass.
Seo, S., Jang, S. S., Almanza, B., Miao, L., & Behnke, C. (2014). The negative spillover
effect of food crises on restaurant firms: Did Jack in the Box really recover from
an E. coli scare? International Journal of Hospitality Management, 39, 107-121.
Thackeray, R., Neiger, B., Smith, A., & Van Wagenen, S. (2012). Adoption and use of
social media among public health departments. BMC Public Health, 12(1), 242.
van Deursen, A. J., & van Dijk, J. A. (2014). The digital divide shifts to differences in
usage. New Media & Society, 16(3), 507-526.
97
van Deursen, A. J. A. M., & van Dijk, J. A. G. M. (2009). Using the Internet: Skill related
problems in users’ online behavior. Interacting with Computers, 21(5), 393-402.
Vorvoreanu, M. (2004). Building and maintaining relationships online: A framework for
analyzing the public relations website experience. Retrieved from ProQuest
Digital Dissertations database.
Vorvoreanu, M. (2006). Online organization–public relationships: An experiencecentered
approach. Public Relations Review, 32(4), 395-401.
Vorvoreanu, M. (2008). Website experience analysis: A new research protocol for
studying relationship building on corporate websites. Journal of Website
Promotion, 3(3-4), 222-249.
Webb, T., Joseph, J., Yardley, L., & Michie, S. (2010). Using the Internet to promote
health behavior change: A systematic review and meta-analysis of the impact of
theoretical basis, use of behavior change techniques, and mode of delivery on
efficacy. Journal of Medical Internet Research, 12(1), e4.
Wilson, T. D. (1981). On user studies and information needs. Journal of Documentation,
37(1), 3-15.
Wilson, T. D. (1999). Models in information behaviour research. Journal of
Documentation, 55(3), 249 - 270.
Witte, K. (1992). Putting the fear back into fear appeals: The extended parallel process
model. Communications Monographs, 59(4), 329-349.
98
Witte, K., & Allen, M. (2000). A meta-analysis of fear appeals: Implications for effective
public health campaigns. Health Education & Behavior, 27(5), 591-615.
Zickuhr, K., & Madden, M. (2012). Older adults and Internet use. Retrieved from
http://www.pewinternet.org/2012/06/06/older-adults-and-internet-use/
CHAPTER 5. STUDY 3: IMPROVING FOOD SAFETY COMMUNICATION ON
THE INTERNET: INFLUENCE OF MEDIA ON COMMUNICATION
OUTCOMES
The following study will be submitted to a peer-reviewed journal and is
included in this non-traditional thesis/dissertation as chapter 5. This article is
written in the APA style format.
5.1 Abstract
Although food safety is important, media’s role in food safety communication has largely
been overlooked. To explore media’s role in influencing food safety communication, this
study applied five methodological procedures to test the relationships among website
characteristics, perceptions towards these characteristics, efficacy perceptions, and
behavioral intentions. It was determined that website characteristics, through efficacy
perceptions, influence consumers’ intentions to use the communicated information. The
results suggest that both information quality and the way information is communicated
have a significant impact on consumers’ behavioral intentions. Website features that
99
directly relate to searchability (e.g. search box and site map) and saliency (specific
information about foods and locations involved) were perceived to be most influential
and should be considered when designing and using their websites.
5.2 Introduction
Consumers can play an important role in the nation’s food safety if they use food safety
information in their food decisions. At the same time, a prerequisite to information usage
is access to information. With the increasing availability and popularity of Internet-based
communication platforms, communicating food safety information online to consumers is
an attractive alternative to traditional medias. Online platforms offer high speed (Tinker
& Fouse, 2009), high scalability (Glasgow et al., 2006), high message fidelity (Snyder,
2001), low costs (Thackeray et al., 2012; Trouten, 2013), and broader reach (Chou et al.,
2009). Nonetheless, challenges remain in leveraging Internet-based platforms in food
safety communication. In particular, consumer concern over information quality can limit
the utility of online platforms. This study is designed to determine how Internet-based
platforms can be used in food safety communication and contribute to the protection of
public health.
5.3 Literature review
Food safety communication offers enormous benefits including the protection of public
health and improved business operations (Almanza et al., 2002; Choi et al., 2011; Jin &
Leslie, 2002; Porucznik & Royal DeLegge, 2013). Traditionally, food safety information
has been communicated through push medias that are passive in message delivery, such
as TV and newspapers. In fact, the most common outlets for food safety information have
100
been newspapers, television, and radio (Almanza et al., 2003). The major drawback in
push communication is that users have few choices about the information they receive
and when, so the information is likely to have low relevancy, resulting in lower
information usage. Additionally, the limited reach of traditional food safety
communications reduces the impact of communication (Charles & Lawrence, 1990;
Dutta-Bergman, 2004). If food safety information was more accessible and the
communication was more engaging, consumers would be more likely to use it (Worsfold,
2006). A recent national survey of 5,000 consumers showed that consumers are more
interested in food safety and transparency throughout the food system (Food Safety
News, 2016). In addition, the Internet is the top choice for consumers seeking food safety
information (Charanza & Naile, 2012).
Literature suggests that, once motivated, information quality, source characteristics, and
media properties all impact the evaluation and later usage of information, and that
information, source, and media characteristics can interact (Barry & Schamber, 1998;
Frighetto & Wolf, 2014; Krikelas, 1983; Kuhlthau, 1991; Park & Lessig, 1981; Tversky
& Kahneman, 1973; Young & Von Seggern, 2001). Specifically, both information and
source characteristics can interact with media characteristics in determining consumer
trust and later communication outcomes. No matter which media consumers use, they
want to get accurate information, and consumers may perceive information from a health
department to be more accurate than information from an individual. Additionally, source
characteristics are often directly related to information characteristics.
While the impacts of source and information have been studied (Gordon, 2003; Kornelis,
De Jonge, Frewer, & Dagevos, 2007), media’s potential contribution in improving
101
communication outcomes has received limited scholarly attention. Overall, within the
context of foodborne illness outbreak communication, the role of media has been largely
overlooked. Research has focused on benefits of new media (Chapman et al., 2014;
Rutsaert et al., 2014), effect of communication and intervention (Mayer & Harrison,
2012; Mitchell et al., 2007; Powell et al., 2011), and user typology (Kuttschreuter et al.,
2014). This lack of research limits scholars’ ability to fully explore ways to improve
usage of Internet-based communication platforms in food safety communication. To
address this issue, this study applies the Extended Parallel Process Model (EPPM) to
examine media’s role in consumers’ food safety information-seeking experience.
Additionally, because of the intercorrelations among media, information, and source
when examining media effect, this study incorporates information and source
characteristics in evaluating media’s influence on communication outcome. In essence,
this study will attempt to determine the media characteristics, and more specifically the
platform features, that contribute to improved foodborne illness outbreak communication
outcomes and why.
Studies have suggested that certain website characteristics such as interactivity, usability,
trustworthiness, information quantity, security, recency, and accuracy (Aladwani &
Palvia, 2002; Coursaris & Sung, 2012; Hanafi et al., 2009; Hsieh, Kuo, Yang, & Lin,
2010; Hsu & Lin, 2016; Klein, 2001; Nelson et al., 2005; Strong, Lee, & Wang, 1997;
Wixom & Todd, 2005) are directly associated with users’ evaluations and thus
satisfaction and usage of a platform. Previous studies (Ma, Almanza, Ghiselli,
Vorvoreanu, & Sydnor, 2016a, 2016b) have suggested that the most relevant
characteristics regarding source, media, and information in foodborne illness outbreak
communication are saliency (information), timeliness (information), trustworthiness
102
(source), searchability (media), usability (media), and interactivity (media). Thus, this
study focuses on three media characteristics (searchability, enhanced usability, and
interactivity), one source characteristic (trustworthiness), and two information
characteristics (timeliness and saliency). In this study, the most commonly preferred
website features from previous studies (Ma et al., 2016a) are used to represent website
media characteristics – specifically, search box and site map for searchability, links to
outside information and links to pictures and videos for linkability, a function to ask
questions and a function to share information for interactivity. See Figure 5-1.
The Extended Parallel Process Model (EPPM) was originally developed to guide message
formation in health communications when risks are involved (Witte, 1992). Witte (1992)
reintegrated affective process with Rogers’s work of cognitive process of fear (1975,
1983), built upon Leventhal’s framework of parallel danger and fear processes (1970),
and introduced EPPM. EPPM suggests that after exposure to a threat (normally in the
form of a fear appeal), an individual develops the perception of threat based on the
severity of the threat and the personal susceptibility to the threat. The more severe the
situation, and the more susceptible the individual is, the more they are motivated to
evaluate the efficacy of the recommended response. Individuals’ perceptions of response
efficacy (how effective the recommended response is in reducing the threat) and self-
efficacy (if they believe they can perform the recommended response) are formed and
combined with their perceived level of threat, thereby leading to their response choice.
There are three possible outcomes. Individuals can ignore the message if the threat is
perceived to be low or irrelevant (so the efficacy appraisal is not even triggered).
Alternatively, individuals can engage in danger control where they process the message
103
and take the recommended actions. This is more likely to occur from a combination of
high threat perception and high efficacy perception. Or individuals can engage in fear
control when they are exposed to a serious threat without a perceived effective response
or they do not believe they can successfully perform the response and avoid the threat. In
other words, a high threat combined with low efficacy will trigger danger control. Since
its introduction, EPPM has been widely adopted to develop effective risk communication
messages, specifically messages that would elicit adaptive behavioral responses (Gore &
Bracken, 2005; McMahan et al., 1998). Outside message design, EPPM’s applications in
assessing attitudes towards an action (Barnett et al., 2009) and predicting behaviors
(Hullett & Witte, 2001; Roberto & Goodall, 2009) supports its predicative power in a
wide range of contexts.
Health communications are increasingly delivered through digital media (Rice & Atkin,
2012; Vance, Howe, & Dellavalle, 2009). The need to examine Internet platforms as new
communication media is pressing. Internet as a media provides unique platform and
information characteristics that could enhance communication effectiveness. These
platform and information characteristics likely lead to perceptions of information quality
– how good the information is – and media usability – how usable the media is (Al-Qeisi
et al., 2014; Aladwani & Palvia, 2002; DeLone & McLean, 1992; Hanafi et al., 2009;
Huang & Benyoucef, 2013; Wixom & Todd, 2005). With encouragement from Kim
Witte (Witte, 2015) to assess the impacts of these unique features, this project used EPPM
with two additional efficacy components – message efficacy and process efficacy (Figure
5-1). Process efficacy was proposed to incorporate media features of Internet platforms,
and message efficacy was added to assess information quality. Message efficacy was
104
included in the proposed model because information quality may be experienced
differently in the digital communication context. For example, timeliness may mean
getting news once per day in traditional media communication, but may be real-time if
the communication is happening online. Efficacy definitions and the statements used to
evaluate consumers’ perceptions of these efficacies are given in Table 5-1.
Figure 5-1. Proposed model representing relationships among EPPM constructs and perceptions
towards website characteristics in foodborne illness outbreak communication Table 5-1. Definitions
of efficacies and statements used to capture each efficacy perception
Efficacy
Definition
Statements
Self-efficacy
A person’s belief that he or
she has the ability to find
needed information (about
foods and restaurants to
avoid).
1.
2.
I feel I am able to find the
information regarding foods and
restaurants to avoid on this website
I have the ability to locate the
information to avoid the foods and
restaurants that are involved in this
outbreak on this website
3.
I am confident that I can find the
information to use to avoid the
foods and restaurants involved in
this outbreak on this website
105
Response efficacy
A person’s belief that using
information found reduces
the risk of getting sick.
1.
I believe that using the information
on this website helps to prevent me
from getting sick.
2.
Using the information that I found
on the website protects me during
this outbreak.
3.
I feel that the use of this website’s
information reduces my risks of
getting sick during this outbreak.
Process efficacy
A person’s belief that the
process of finding needed
information is easily done.
1.
2.
I think getting to the information I
need is easy on this website I feel
the information I need is easily
accessible on this website.
3.
I feel it is easy for me to find what
I am looking for on this website
Message efficacy
A person’s belief that the
information found is of high
quality.
1.
2.
I feel the information found on this
website is of high quality. I think I
find high-quality information on
this website.
3.
I believe that the quality of
information found on this website
is high.
In summary, this study examined how various characteristics of Internet platforms
influence the perceptions of efficacy in providing food safety information, and the
communication outcome. Results were expected to offer practical suggestions to improve
food safety communication and advance the theoretical understandings of users’ food
safety information-seeking behaviors. The following research questions were proposed:
1. How will different Internet platform characteristics (media, information, and source
characteristics) contribute to improving food safety communication outcomes
(behavioral intentions)?
2. When confronted with a threat, will a higher level of perceived efficacy (including
process and message efficacy) improve communication outcomes?
3. Is EPPM useful in understanding food safety information-seeking behaviors?
106
Hypotheses emerging from these research questions were:
1. Website characteristics (source, information, and media) affect users’ perceptions of
the website.
2. Users’ perceptions related to information quality (trustworthiness, timeliness, and
saliency) positively affect message efficacy in experiencing the website.
3. Users’ perceptions related to media usability (searchability, linkability, and
interactivity) positively affect process efficacy in experiencing the website.
4. Higher perceived message efficacy leads to stronger information usage intention.
5. Higher perceived process efficacy leads to stronger information usage intention.
EPPM predicts that when confronted with a high level of threat, individuals with higher
perceived levels of efficacy are more likely to engage in protective behaviors (Witte,
1994). Thus, it is proposed:
6. Higher perceived self-efficacy leads to stronger information usage intention.
7. Higher perceived response efficacy leads to stronger information usage intention.
For individuals, perceived easiness of process in locating needed information will not
only influence individuals’ information usage intention but also individuals’ perceived
ability to locate the information (Davis, 1989; Legris, Ingham, & Collerette, 2003;
O'Reilly, 1982; Venkatesh, Morris, Davis, & Davis, 2003). Thus, it is proposed:
8. Higher process efficacy perception will lead to improved perceived self-efficacy.
Witte and Allen (2000), in a meta-analysis, have suggested that self-efficacy, to a certain
extent, incorporates the idea of “easiness of the process.” In other words, how individuals
107
perceive their ability to find the information will impact how process efficacy influences
information usage intention. Thus, this study further proposes:
9. Self-efficacy mediates the relationships between process efficacy and consumers’
information usage intention.
Information quality will also directly impact consumers’ intended information usage as
well as their perceived “usefulness” of the information (Davis, 1989; Jingjun, Benbasat,
& Cenfetelli, 2013; Legris et al., 2003; Venkatesh et al., 2003). As such, it is
hypothesized:
10. Higher message efficacy perception will lead to higher perceived response efficacy.
Additionally, even if the information is good, if it is not useful, individuals may still not
use it. Sometimes, information of lower quality can still be useful (Miller, 1996). This
phenomena exists in food safety communication as well, the usefulness of the
information can influence the impact information quality has on behavioral intention.
Thus, it was further proposed:
11. Response efficacy mediates the relationships between message efficacy and
consumers’ information usage intention.
Users generally experience a website as a whole (including information, source, and
media characteristics) and form their perceptions and evaluations based on their overall
experience. Thus it is hypothesized:
12. Process efficacy and message efficacy covary.
108
Lastly, in traditional applications of EPPM, self- and response efficacy are often
combined to calculate an “efficacy score” (Witte, 1996), so self- and response efficacy
are expected to be correlated.
13. Self-efficacy and response efficacy covary.
5.4 Method
5.4.1 Measurement
Because consumers may have varying levels of urgency and continuity for different types
of food safety information (restaurant inspection results, food recalls, and foodborne
illness outbreaks), this study focused only on foodborne illness outbreak information
(FBI). Measures of threat (susceptibility and severity) and efficacy (self-efficacy and
response efficacy) were adopted from Witte’s RBD scale (1996) and modified to suit FBI
communication. Based on previous research (Ma et al., 2016a, 2016b), three
characteristics pertaining to perceived information quality (timeliness, trustworthiness,
and accuracy and/or precise detailed information) were assessed. Additionally, three
characteristics pertaining to perceived ability to use the media, searchability, enhanced
usability, and interactivity, were found to be important (Ma et al., 2016a, 2016b) and thus
included in this study. Six specific features that represented searchability, linkability, and
interactivity from the previous research study were used to assess perceptions. These
included a search box and site map (searchability), functions to share the information
with others and allow users to ask questions and provide feedback (interactivity), and
links to additional information and to pictures or video (linkability). Behavioral intentions
in this study included intentions to use the information on the website in making dining
109
choices in both the short-term and in the long-term. All statements and questions were
measured on a 7-point Likert scale.
5.4.2 Data collection
Two stages of data collections were implemented in this study: First, the link between
users’ perceptions and the presence of different website features was evaluated in order to
determine whether the proposed website characteristics created perceptions of high
information quality and improved usage experience; second, the hypotheses were tested
to examine the ways website features impact communication outcomes. At both stages,
surveys were distributed through an online company, Amazon Mechanical Turk (mTurk),
to randomly select online participants in the United States.
During the first stage, questions were asked to map out how different levels of website
features affected consumers’ perceptions. For example, participants were asked to rate
their perceived ability to search for information on a website when different levels of
search functions (e.g. presence of a search box vs. absence of a search box) were offered
on the website. This information is also used in the 2nd stage of data collection to set up
the scenarios.
Additionally, the perception of (hypothetical) outbreaks with different severity levels was
assessed in stage 1 for their effect on participants’ perceptions of threat. Because the
number of people sickened and the number hospitalized are often used to characterize an
outbreak, this study presented five levels of severity with varying numbers of sick (2, 25,
and 100) and hospitalized cases (0, 4, and 17). Two people is the minimum required for
an event to be called an outbreak (Delone & McLean, 2003). Numbers of people
110
hospitalized was calculated at 17%, the average percentage of victims hospitalized during
many multi-state salmonella outbreaks in the past three years (CDC, 2000). Salmonellosis
is responsible for the majority of hospitalizations among foodborne pathogens (CDC,
2016e).
After validating the importance of the specific website features to the formation of user
perceptions in stage one, a self-administered questionnaire was developed for stage two.
Following a description of an FBI outbreak, questions were asked regarding threat
perceptions (susceptibility and severity). Next, ten scenarios described available website
features and information content (each participant received only one scenario) were
created. One scenario had all characteristics, 6 scenarios each contained one characteristic
(searchability, linkability, interactivity, timeliness, trustworthiness, and saliency), 2
scenarios were used to capture process efficacy related characteristics (searchability,
linkability, interactivity, and timeliness) and message efficacy related characteristics
(timeliness, trustworthiness, and saliency), and one scenario where all the desired
characteristics were absent.
Following were questions assessed perceptions towards website characteristics, efficacy
perceptions (self, response, process, and message efficacy), behavioral intention, and
demographics.
5.4.3 Data analysis
Stage 1. A total of 209 responses were collected. After eliminating responses with the
same IP address, same mTurk worker ID, high percentage of incomplete answers, or took
less than 5 minutes to complete, a total of 195 responses were used for the data analysis.
111
ANOVA tests were used to verify the links between website characteristics and users’
perceptions and to facilitate the setup of scenarios used in Stage 2. Data analyses were
conducted using SPSS version 23.
Stage 2. Two samples of 211 and 550 were collected (one sample was used for
Exploratory Factor Analysis and the other sample was used for Confirmatory Factor
Analysis and Structural Equation Modeling). After using the same data cleaning
procedure, two samples of 198 and 511 responses were used in the data analysis.
Structural Equation Modeling (SEM) was used to test the relationships among proposed
constructs and to understand the mechanism behind the influence of website
characteristics on communication outcomes. A two-step approach was adopted to test the
measurement model and structural model to ensure the quality of measures (Anderson &
Gerbing, 1988). For the first step, confirmatory factor analyses (CFA) were performed
and the reliability and validity of the measurement models were examined. In the second
step, SEM with a maximum likelihood estimate was performed to identify the
relationships among the proposed constructs. Goodness-of-fit (GOF) indices, including
, root mean square error of approximation (RMSEA), absolute fit indices and normed
fit index (NFI), Tucker-Lewis index (TLI), and comparative fit index (CFI), were checked
and established to be within the acceptable levels. To test mediation effect, procedures
using the Bootstrapping method were carried out (Gunzler, Chen, Wu, & Zhang, 2013;
Shrout & Bolger, 2002). All data analyses were finished using SPSS version 23 and
AMOS version 23.
112
5.5 Results
5.5.1 Stage one
There were slightly more females than males, and about 67% of participants were 20 to
39 years old (Table 5-2). About 46% of respondents reported that they had a bachelor’s
degree or higher and 68.2% had no children. Participants were from all over the nation.
Table 5-2. Profiles of Respondents
Stage 1 of data collection
Stage 2 of data collection
ANVOA sample
(n=195*)
EFA sample (n=198*)
SEM sample (n=511*)
Characteristics
n
%
n
%
n
%
Gender
Male
95
48.7 (49.2)
78
39.0 (49.2)
203
39.7 (49.2)
Female
100
51.3 (50.2)
122
61.0 (50.2)
308
60.3 (50.2)
Age
18 – 29
75
38.5 (18.9)
55
27.5 (18.9)
145
28.4 (18.9)
30 – 39
59
30.3 (17.8)
69
34.5 (17.8)
158
30.9 (17.8)
40 – 49
26
13.3 (19.3)
35
17.5 (19.3)
93
18.2 (19.3)
50 – 59
26
13.3 (18.6)
31
15.5 (18.6)
66
12.9 (18.6)
60 +
9
4.6 (25.3)
10
5.0 (25.3)
49
9.6 (25.3)
Household with Children
No Children
133
68.2
110
55
309
60.5
Children under 6 years old
19
9.7
26
13
67
13.1
Children 6 years old and
over
39
20
47
23.5
115
22.5
Others
4
2.1
17
8.5
19
3.9
Education
Less than Bachelor’s Degree
106
54.4 (70.7)
94
47.0 (70.7)
222
43.8 (70.7)
Bachelor’s Degree
71
36.4 (18.9)
72
36.0 (18.9)
199
38.9 (18.9)
Higher than Bachelor’s
Degree
18
9.2 (10.4)
34
17.0 (10.4)
88
17.2 (10.4)
113
Residential Area
New England
5
2.6 (4.7)
8
4.1 (4.7)
17
3.4 (4.7)
Mid Atlantic
26
13.4 (13.3)
20
10.3 (13.3)
81
16.0 (13.3)
East North Central
25
12.9 (15.2)
40
20.6 (15.2)
70
13.9 (15.2)
West North Central
5
2.6 (6.6)
13
6.7 (6.6)
34
6.7 (6.6)
South Atlantic
48
24.7 (19.3)
48
24.7 (19.3)
120
23.8 (19.3)
East South Central
17
8.8 (6.0)
6
3.1 (6.0)
22
4.4 (6.0)
West South Central
24
12.4 (11.7)
18
9.3 (11.7)
53
10.5 (11.7)
Mountain
19
9.8 (7.2)
12
6.2 (7.2)
34
6.7 (7.2)
Pacific 25 12.9 (16.1)
29
14.9 (16.1)
74
14.7 (16.1)
Note: numbers in parentheses are 2012 U.S. census data
* Not every respondents answered all the questions
Results showed that there are significant differences among users’ perceptions of
searchability, linkability, interactivity, timeliness, trustworthiness, and saliency when the
different levels of website characteristics were presented (Table 5-3). Hypothesis one was
therefore supported. Results from stage were also used for minor wording changes in
stage two to more accurately reflect users’ experience.
Table 5-3. One-way analysis of variance for perceptions towards website features/characteristics
(n=195)
Perception
Features/ characteristics
t/F
p
Searchability
Search box and site map
624.665
<0.001
Linkability
Links to additional information and links to
picture or video
392.512
<0.001
Interactivity
Function to ask questions and provide feedback
and function to share with others
497.8
<0.001
Timeliness
Update frequency
192.192
<0.001
Saliency
Names of restaurant and foods specifically
mentioned
160.589
<0.001
Trustworthiness
Health department website
46.037
<0.001
5.5.2 Stage two
Descriptive information about participants is presented in Table 5-2. The two samples
were similar. Both samples included more females than males, and a little more than 60%
of participants were 20 to 39 years old. More than half of the respondents reported that
114
they had a bachelor’s degree or higher, and more than 50% had no children in both
samples. The participants were from all over the nation.
Manipulation check
Threat. The averages for participants’ susceptibility and severity perceptions were 4.54
and 5.31 (out of 7) respectively, indicating that the scenarios generated a perception of
threat. Furthermore, 86% of participants indicated that they would look for more
information about the foodborne illness outbreak, showing that the scenario successfully
created the motivation to seek additional information.
Perceptions towards website characteristics. A series of t-tests indicated that participants
perceived different levels (e.g. high vs. low and adequate vs. inadequate) of searchability,
linkability, interactivity, timeliness, saliency, and trustworthiness when different website
characteristics were presented (Table 5-4) and verified that the links between website
characteristics and perceptions towards these characteristics worked as expected and thus
demonstrated the manipulations were successful.
Table 5-4. Results of t-test in the manipulation check (n=709*)
High
Low
Significance
Mean
SD
Mean
SD
t-value
Trustworthiness
6.07
0.90
2.76
1.40
17.48***
Searchability
6.01
0.98
2.83
1.32
17.01***
Linkability
6.04
0.97
2.19
1.35
20.27***
Interactivity
5.79
1.11
2.13
1.38
18.09***
Timeliness
5.95
1.14
2.54
1.30
17.32***
Saliency
6.05
0.94
2.47
1.38
18.79***
*** p < 0.001
Note: Each participant received only 1 scenario.
115
* The two samples of 198 and 511 were combined here for the manipulation check.
EFA
Exploratory factor analysis (EFA) was conducted to examine the underlying factorial
structures. Thirteen factors were extracted based on the theoretical underpinnings
discussed before (Appendix 1). The proposed factor structure (Figure 5-1) was supported,
with the exception of two saliency items (saliency2 and saliency1) cross loading on
timeliness, and two linkability (linkability1 and linkability3) items cross loading on
searchability. KMO and Bartlett’s test of Sphericity were examined to access the
adequacy of sample. The KMO statistic of .957 for this study fell in the range of being
excellent (Hair, 2010); therefore, the sample size of the current study was considered
adequate for factor analysis. Overall, the 13 factors explained 86.61% of the variance;
and the internal consistency of items within each construct ranged from 0.770 to 0.975,
indicating that the reliability of the measurements was satisfactory, thus it was used later
for the CFA analysis.
CFA
Following EFA, a confirmatory factor analysis (CFA) was performed to assess the
measurement model as recommended in the two-step approach to SEM (Kaiser, 1974).
See Table 5-5. After examining the goodness-of-fit indices for the measurement model, it
appears that the measurement model reflected a reasonably good fit to the data
(2=1387.750, df =587, p < 0.001, 2/df = 2.364, NFI = 0.950, TLI = 0.965, CFI = 0.971,
IFI = 0.971, RMSEA = 0.052).
116
Reliability and validity. The reliability of the measurement items was assessed using
Cronbach’s α. The value of Cronbach’s α (internal consistency) for the three constructs
ranged from 0.770 to 0.975, exceeding the minimum required value of 0.7 (Hair, 2010).
Thus, the level of internal consistency for each construct was considered acceptable.
Additionally, Construct composite reliability ranged from 0.850 to 0.959, all exceeded the
cutoff value of 0.7 (Hair, 2010), demonstrating internal consistency as well.
To evaluate convergent validity, standardized factor loadings for all measurement items
and the average variance extracted (AVE) were estimated. All measurement items had
standardized loadings of .5 or higher (ranging from 0.775 to 0.960) and these loadings
were all significant at alpha level of 0.001. Further, all AVEs for the constructs (ranging
from 0.653 to 0.908) exceeded the recommended hurdle of .5 (Fornell & Larcker, 1981).
These results provided evidence of acceptable convergent validity – the variance
extracted by measurement items was greater than variance due to measurement error
(Fornell & Larcker, 1981).
Discriminant validity was reviewed by comparing squared correlations between the
constructs with the AVE. See Appendix 2 for correlation matric between constructs. In
checking discriminant validity, it was found that two pairs of variables were highly
correlated (where at least one of the two AVEs did not exceed the squared correlation).
They included searchability and linkability, and searchability and interactivity. This
reflects similar problems as observed in EFA, but was somewhat expected as consumers
experience the website as a whole and certain features are not perceived alone, separate
from other related features. Additionally, previous research (Ma et al., 2016a) suggested
that consumers experience functions or features linked to various perceptions differently.
117
For example, consumers used search boxes (linked to perception of searchability) to
locate the information more quickly, but used links (connected to perception of
linkability) to get additional information. As such, the highly correlated perceptions
towards website characteristics were left in the model and path coefficients were
examined later.
Table 5-5. Results of confirmatory factor analysis for measurement model
Variables
Standardized
loading
P-value
Cronbach's α
Composite
reliability
AVE
Susceptibility
0.770
0.850
0.653
I am at risk for getting sick
0.824
< 0.001
It is likely that I will get sick
0.775
< 0.001
It is possible that I will get sick
0.826
< 0.001
Severity
0.903
0.901
0.753
I believe this foodborne illness outbreak is severe
0.855
< 0.001
This foodborne illness outbreak sounds serious to me
0.883
< 0.001
I believe that this foodborne illness outbreak is significant
0.865
< 0.001
Trustworthiness
0.969
0.959
0.886
The trustworthiness of this website
0.915
< 0.001
The reliability of this website
0.951
< 0.001
Your trust of this website
0.956
< 0.001
Searchability
0.960
0.945
0.852
Information searching
0.914
< 0.001
The ability to look for information on this website
0.932
< 0.001
This website’s “searchability”
0.923
< 0.001
Linkability
0.956
0.952
0.869
This website’s ability to connect you with other sources of
information
0.943
< 0.001
This website’s ability to provide you with information from
other experts
0.921
< 0.001
This website’s use of links to provide more information
0.935
< 0.001
Table 5-5 continued
Variables
Standardized
loading
P-value
Cronbach's α
Composite
reliability
AVE
Interactivity
0.954
0.967
0.908
Ability to interact on this website
0.927
< 0.001
Ability to communicate with others on this website
0.941
< 0.001
Ability to exchange information on this website
0.961
< 0.001
Timeliness
0.956
0.950
0.862
Frequency of the updates
0.922
< 0.001
The timeliness of this website
0.943
< 0.001
Currency of the information on this website
0.921
< 0.001
Saliency
0.962
0.949
0.862
Comprehensiveness of information on this website
0.93
< 0.001
Adequate amount of information on this website
0.917
< 0.001
Adequate amount of detail on this website
0.937
< 0.001
Self-efficacy
0.955
0.944
0.850
I am able to find the information regarding what foods and
restaurants to avoid
0.918
< 0.001
I have the ability to locate the information I need
0.935
< 0.001
I am confident that I can find information to avoid the foods
and restaurants involved in this outbreak
0.914
< 0.001
Table 5-5 continued
Variables
Standardized
loading
P-value
Cronbach's α
Composite
reliability
AVE
Response efficacy
0.947
0.939
0.838
I believe that using the information on this website prevents
me from getting sick
0.866
< 0.001
The use of this website’s information will help to protect me
during this outbreak
0.94
< 0.001
I feel that the use of this website’s information reduces my risk
of getting sick during this outbreak
0.938
< 0.001
Process efficacy
0.975
0.959
0.885
I think getting to the information I need is easy on this
website
0.92
< 0.001
I feel the information I need is easily accessible on this
website.
0.951
< 0.001
I feel it is easy for me to find what I am looking for on this
website
0.955
< 0.001
Message efficacy
0.965
0.953
0.872
The information on this website appears to be good.
0.924
< 0.001
I think this website has good quality information
0.928
< 0.001
I believe the quality of information on this website is high
0.946
< 0.001
Intention
0.917
0.917
0.848
I will use information on this website to make dining out
decisions during this outbreak
0.937
< 0.001
I will use information on this website to make dining out
decisions in the future
0.904
< 0.001
Model fit: 2=1387.750, df =587, p < 0.001, 2/df = 2.364, NFI = 0.950, TLI = 0.965, CFI = 0.971, IFI = 0.971, RMSEA =
0.052.
128
Structural model and hypotheses testing
The proposed model shown in Figure 5-1 was estimated using Structural Equation
Modeling (SEM) with maximum likelihood estimation to investigate the relationship
among the proposed constructs — namely, perceptions of threat (susceptibility and
severity), perceptions towards website characteristics (trustworthiness, searchability,
linkability, interactivity, timeliness, and saliency), efficacy perceptions (self-efficacy,
response efficacy, process efficacy, and message efficacy), and behavioral intentions. The
goodness-of-fit statistics for the structural model indicated that the proposed model
reasonably fit the data (χ2 = 2556.657, df = 649, p < 0.001, χ2/df = 3.939, NFI = .908, TLI
= .923, CFI = .929, IFI = .929, RMSEA = .077). See Table 5-6 for the final structural
model results and Figure 5-2 for a graphic presentation of the final structural model
estimations with standardized path coefficients.
Table 5-6. Structural parameter estimates and fit indices (n=511)
Path
Standardized
estimate
t-statistic
P-value
Relationship
Timeliness → Message Efficacy
-0.181
-3.369
0.001***
Significant
Saliency → Message Efficacy
0.702
10.127
0.001***
Significant
Trustworthiness → Message Efficacy
0.391
9.314
0.001***
Significant
Searchability → Process Efficacy
0.847
5.903
0.001***
Significant
Interactivity → Process Efficacy
-0.071
-0.817
0.414
Not significant
Linkability → Process Efficacy
0.018
0.152
0.879
Not significant
Process Efficacy→ Self Efficacy
0.87
29.312
0.001***
Significant
Message Efficacy → Response Efficacy
0.893
29.355
0.001***
Significant
Process Efficacy → Intention
0.044
0.608
0.543
Not significant
Message Efficacy → Intention
0.342
4.08
0.001***
Significant
Self-Efficacy → Intention
0.579
7.606
0.001***
Significant
Response Efficacy → Intention
0.182
2.125
0.034*
Significant
Susceptibility → Intention
0.019
0.453
0.65
Not significant
Severity → Intention
-0.013
-0.319
0.75
Not significant
Message Efficacy ↔ Process Efficacy
0.574
11.704
0.001***
Significant
Self Efficacy ↔ Response Efficacy
0.226
5.614
0.001***
Significant
Goodness-of-fits statistics
Structural Model
Cut-off value
129
Chi-square = 2556.657
N/A
Normed Chi-square = 3.939
1.0-5.0
NFI = .908
>.90
TLI = .923
>.90
CFI = .929
>.90
IFI = .929
>.90
RMSEA = .077
.05-.08 Mediocre fit
<.05 Good fit
Note: *p<0.05,**p<0.01,***p<0.001
Figure 5-2. Results of the structural model
The results of the structural model indicated that perceptions of trustworthiness (β =
0.391, p < .001), timeliness (β = -0.181, p < .001), and saliency (β = 0.702, p < .001)
130
would all significantly impact consumers’ evaluation of message efficacy. However, it is
noticed that timeliness negatively impacts consumer perceived message efficacy.
Hypothesis 2 was partially supported. Among the three, saliency was identified as the
most significant factor that stimulates the perception of message efficacy or that the
information is of high quality; 49.28% of variance of the message efficacy construct can
be explained by information saliency. This clearly shows the importance of information
saliency. On the other hand, it was surprising to see that timeliness had a negative
coefficient estimate, the opposite of what had been hypothesized.
Regarding process efficacy, out of the three proposed and tested perceptions, only
searchability had a significant impact, but this impact was considered major as the path
coefficient was 0.847 (p < .001). Hypothesis 3 was partially supported. This result further
suggested that 71.74% of the variance in process efficacy could be accounted for by the
perception of searchability, demonstrating the significance of searchability. Neither
linkability (β = 0.018, p = 0.879) nor interactivity (β = -0.071, p = 0.414) was a
significant predictor of process efficacy. It was surprising to observe that interactivity had
a negative influence (even though not significant) on perception of process efficacy.
In terms of relationships among the efficacy perceptions, message efficacy was a strong
predictor of response efficacy (β = 0.893, p < .001) while process efficacy was also rather
predictive of self-efficacy (β = 0.870, p < .001). When examining the influence of
efficacy perceptions on behavioral intention, both self-efficacy (β = 0.579, p < .001) and
response efficacy (β = 0.181, p < .05) had a direct impact on behavioral intention.
Hypotheses 6 and 7 were supported. Additionally mediation analyses applying the
bootstrapping method were conducted (Gunzler et al., 2013). See Table 5-7. It is seen that
process efficacy had a direct impact on behavioral intention (β = 0.461, p < 0.001), but
131
self-efficacy fully mediates this impact. In other words, the relationship between process
efficacy and information usage intention drops and becomes insignificant when
selfefficacy is included as a mediator. Hypotheses 5, 10, and 11 were supported. Message
also had a direct impact on behavioral intention (β = 0.624, p < 0.001) when response
efficacy was not included as a mediator. When response efficacy was tested as a mediator,
response efficacy partially mediates the relationship between message efficacy and
behavioral intention. This means that response efficacy helps to explain some, but not all
of the relationship between message efficacy and behavioral intention. Hypotheses 4, 12,
and 13 were supported. See Appendix 3 for a summary of results of hypotheses testing.
Lastly, it was determined that process and message efficacy covaried (β = 0.574, p <
.001), as did response and self-efficacy (β = 0.226, p < .001). Hypotheses 14 and 15 were
supported. This further indicated that users experience the website as a whole, and even
when they form different evaluations of the website (for example, perception regarding
information quality – message efficacy, versus perception towards media usability –
process efficacy), such evaluations were not entirely separate from the rest of the
experience.
Table 5-7. Mediation analyses (n=511)
Path
Direct without
mediator
Standardized
estimate
P-value
Direct with
mediator
Standardized
estimate
P-value
Indirect
Standardized
estimate
P-value
Result
Process Efficacy
→ Intention
0.461 (0.028)
0.001***
0.044 (0.062)
0.543
0.506 (0.069)
0.001***
Full
mediation
Process Efficacy→
Self Efficacy
0.870 (0.030)
0.001***
Self-Efficacy →
Intention
0.579 (0.063)
0.001***
Message Efficacy
→ Intention
0.624 (0.031)
0.001***
0.342 (0.074)
0.001***
0.161 (0.08)
0.045*
Partial
mediation
Message Efficacy
→ Response
Efficacy
0.893 (0.029)
0.001***
Response Efficacy
→ Intention
0.182 (0.079)
0.034*
Note: *p<0.05,**p<0.01,***p<0.001
132
5.6 Discussion
As consumer interest in food safety increases (Grewal, Cote, & Baumgartner, 2004;
Jöreskog, 1999; Mansfield & Helms, 1982), the need to more effectively communicate
food safety information is pressing. In the attempt to explore the role of Internet-based
platforms, and understand whether and how different Internet platform characteristics
(media, information, and source characteristics) contribute to improving food safety
communication outcomes (behavioral intentions), this study first proposed a model using
different perceptions to organize and make sense out of consumers’ experience using food
safety communication websites. This study further investigated the impacts of identified
perceptions on consumers’ information usage intention. Lastly, this study also assessed
EPPM’s utility in studying food safety communication.
Results from this study supported a new theoretical model. This model expands on
previous approaches in three ways: a) by integrating media characteristics in evaluating
and predicting communication outcome; b) by incorporating additional efficacy
perceptions relevant to individual information seeking behaviors beyond what’s included
in the EPPM; and 3) by establishing relationships among efficacy perceptions –
particularly the discovery of full and partial mediation effect; 4) by linking physical
attributes of a communication media (website in this case) to users’ perceptions.
To achieve the study objectives, five methodological procedures were applied. The results
of Stage 1 data analyses verified the proposed relationships between website
characteristics (e.g. search box) and consumer perceptions (e.g. searchability). The EFA
results suggested that the factorial structure proposed in Figure 5-1 was supported. CFA
verified the reliability of the measurement model. Through SEM analyses, this study
133
discovered that searchability was the most influential factor for process efficacy, and
saliency was the most important information attribute that impacts message efficacy.
Additionally, both process efficacy and message efficacy had a direct relationship with
behavioral intention. Utilizing mediation analyses, it was discovered that self-efficacy
medicated process efficacy’s impact on behavioral intention while response efficacy
medicated message efficacy’s influence on behavioral intention. Lastly, message and
process efficacy as well as self- and response efficacy were correlated.
The results taken as a whole suggest that consumers experience various website
characteristics (information, source, and media) and form perceptions towards these
characteristics; these perceptions then lead to evaluations of efficacies and eventually
behavioral intention. These findings showcase the ways media can influence the
communication outcomes in mediated communication situations – media, or where the
information is communicated, matters. This study not only presents the importance of
media (as process efficacy is seen to directly predict behavioral intention), but also
identifies the essential features that entities interested in communicating food safety
information should consider. For example, having a functional search box and a site map
improves user experience in looking for information (making the process easier), and
thus increases their intention to use the information. If consumers are informed and
taking actions to protect themselves, the losses and impacts of a foodborne illness
outbreak might be greatly reduced. Additionally, supplying high quality information
should still be the highest priority for entities interested in communicating foodborne
illness outbreak information as message efficacy was found to have a strong direct impact
on consumers’ information usage intention.
134
This study discovered that self-efficacy fully mediates the relationship between process
efficacy and information usage intention. This means that self- efficacy explains why
there’s a relationship between process efficacy and behavioral intention. Improvements in
the user experience in providing features that make the process of locating needed
information easy will help to increase consumers’ usage of information. Perceived
easiness of use increases one’s belief in his or her ability to find needed information
(about foods and restaurants to avoid) and thus makes it individuals more likely to use the
information communicated. This has very important theoretical and practical
implications.
Theoretically, self-efficacy presents great explaining power in understanding how media
features impact consumers’ intended information usage. This allows future research to
easily incorporate and test additional media features and extend the results of this study.
Further, self-efficacy is definitely at core in understanding consumers’ information
seeking behaviors and thus scholars may want to include this variable in future
explorations, particularly in searching for the reasons behind consumer information
behaviors. Lastly, this research also highlights the importance of user-center research in
food safety communication. Without an understanding of how consumers form
perception regarding their ability to look for information and factors affecting this
process, it is impossible to successfully carry out effective communications.
On a more practical note, users’ perceptions of the functionalities or features on the
website are important. If they are not perceived as helpful in locating the information,
they will not contribute to more information usage by consumers. This can happen when
135
unfamiliar features are provided in an attempt to reduce users’ information seeking
efforts. For example, previous research revealed that when consumers were provided with
a site map that organized information on an unfamiliar topic (e.g. by pathogen names),
they found it harder to locate the information and tended to not use the information (Ma
et al., 2016a). The disconnect between functionalities and consumer perception can also
happen when features are perceived to be irrelevant to information seeking. For example,
the provision of two search boxes were observed to confuse consumers and have an
inverse impact on consumers’ information usage (Ma et al., 2016a). With these
suggestions in mind, government and health agencies interested in communicating food
safety information online should provide familiar features (e.g. those offered on other
commonly used websites) for best communication of information.
Self-efficacy can also be influenced by one’s existing knowledge and familiarity with a
topic as well as one’s interest in food safety. For example, for individuals who know food
safety very well, they would know what keywords to put in and thus more easily find
what they are looking for; a search box would be very useful in this case. But for a
consumer who has very limited knowledge about foodborne illness outbreaks, a search
box may not be as useful. In this sense, providing features in an attempt to improve
consumer information seeking experience may work better for individuals who are more
interested in and familiar with food safety. In this sense, adapting a “universal design”
mindset – designing websites to be inclusive and easy to use for everyone regardless of
their knowledge of food safety can be helpful. For example, instead of listing and
organizing outbreaks around pathogens that caused them, using food product or location
to organize information will improve the experience for all consumers including those
that are not as familiar with food safety topics. Also, using keywords that are generally
136
used by consumers in searching to index information will also help. In fact, in study 2, it
was found that the keywords majority of the participants used are not recognized by the
search function and yielded no relevant results. In the future, utilizing big data analytics
and identifying the most commonly used keywords by consumers when searching for
information about foodborne illness outbreaks will greatly help to direct the
communication efforts of government and health agencies.
Response efficacy is found to mediate the relationship between message efficacy and
behavioral intention. This finding is particularly interesting in that it allows the
understanding of why information quality impacts consumer information usage intention
– through perceived usefulness. In other words, good information will only be used when
it is perceived useful by consumers.
Research suggests that the difference in “good (high quality) information” vs. “useful
information” is important to consumers. Quality of the information appears to be an
evaluation based on information and source characteristics while the usefulness of
information appears to be an evaluation based on personal experiences and situations.
The following examples might better illustrate this difference. A piece of information that
satisfies the criteria of being good (e.g. specific, timely, and trustworthy) tells you that
the best way for you to wash your hands is to scrub with iodine soap for 5 minutes in a
sterilized environment similar to what a surgeon does. This information is not very useful
for consumers as it is probably impossible for consumers to do, as well as unnecessary in
the home environment. Alternatively, “poor” quality information might still be viewed as
very useful. For example, you found out from the health department that lettuce in your
area is contaminated, but it is not clear which kind of lettuce is specifically linked to this
137
contamination. With more specific (high quality) information, you probably would not
need to avoid all lettuce varieties, but this piece of information, despite its “poor quality”,
can still help you protect yourself if you avoid eating lettuce. Because of this, government
and health agencies should always keep in mind the utility of the information they want
to communicate. The provision of actionable recommendations or tips that are easy to do
and practical is a good practice (e.g. recommending safe cooking temperatures for
potentially contaminated foods). All in all, a trustworthy source’s provision of high
quality information that is very specific and timely is very important; but if the
information is not practical or requires too much effort to implement, consumers still will
not use the information.
Additionally, it was found that consumers experience the website as a whole, and their
perceptions are complex as message and process efficacy covary. In other words, good
user experience (process efficacy) on a website will likely increase consumers’
perceptions of information quality (message efficacy), and vice versa. This showcases
the importance of improving users’ information-seeking experience while maintaining the
information quality. The traditional belief that if the information is good, users will find it
and use it is proved to be inaccurate. This does not diminish the importance of
information quality. As evidenced, message efficacy is the strongest direct predictor of
information usage intention. This finding aligns with the discoveries of earlier research
(Ma et al., 2016b); information quality is consumers’ highest priority in seeking
foodborne illness outbreak information. It is therefore important to maintain or improve
information quality – focusing on three dimensions, saliency, trustworthiness, and
timeliness, while supplying features that would reduce users’ information-seeking efforts
and make the process easier.
138
Taking a more focused view, saliency, timeliness and trustworthiness are all important in
consumers’ evaluations of information quality (message efficacy), but saliency is clearly
the most influential. In the past, the focus for improving food safety communication,
particularly involving foodborne illness outbreaks, has been on timely updates (Food
Safety News, 2016). But results of this study suggested that being specific (saliency) is
more important than being timely (timeliness). When information is available, naming the
specific foods and locations involved is particularly helpful; after all, consumers cannot
make informed decisions if all they know is that an outbreak is occurring.
It was very surprising to observe that timeliness had a negative impact on message
efficacy. One possible explanation for the negative relationship is that daily and monthly
updates were used to frame timeliness of the information in the scenario. Consumers may
still want to see timely information, but daily updates may be too frequent. If updates are
posted daily, additional effort is required to keep track of information. Also, daily updates
may give consumers the perception of constantly changing information and reduce the
perceived trustworthiness of the information. Thus, it is recommended that government
and health agencies only update when needed (e.g. a new location or food is identified)
and not update for the sake of updating. It seems that consumers want food safety
communication to tell them what to do and not change. A website could be refreshed with
the date the information was last updated (which could be every day during a serious
foodborne illness outbreak) to show that the website is attended, but not changing the
content when there is no new information appears to be a good practice. An RSS feed,
which allows consumers who sign up to receive updates when they occur (rather than go
to a website to check for frequent updates), might also be employed.
139
Looking at process efficacy more closely, it is shown that searchability is the only
significant factor influencing consumer evaluation on ease of the process. This result
highlights the central role searchability plays, especially with 71.74% variance explained
solely by searchability. This means that when designing websites, a functional search box
and a site map are considered must-haves. Also, features (such as keyword indexing, texts
from videos, and pop-up boxes suggesting relevant information) that enhance consumers’
searching experience and allow them to locate the needed information more quickly
while reducing the effort required should also be considered and incorporated. On the
other hand, it is not wise to conclude that searchability is the only important factor; in
fact, approximately 28.26% of the variance might be explained by other process efficacy
determinants. More studies are needed to verify and expand our understanding.
In the determinants of process efficacy, interactivity behaves, even though not
significantly, contrary to what has been hypothesized – it is predicted to have a positive
impact on process efficacy. This may be because consumers do not value the
opportunities to interact, as was suggested in earlier research (Ma et al., 2016b). Another
possibility is that consumers found interactive features to be inefficient and interfere with
their process of seeking information. For example, if interactive features are offered,
people can post questions and receive answers on another page or be directed to a
different website or social media page for information. This may interfere with their main
goal of locating the needed information to protect themselves, and reduce the usability of
the website (Ma et al., 2016a). Aside from weeding through information, consumers
appear to not like obtaining information from others – such information might be
perceived as less trustworthy. Additionally, interactive features can be unfamiliar and thus
confusing to consumers. For example, earlier research (Ma et al., 2016a) noted that
140
consumers found the functionality of different ‘buttons” confusing – sometimes it was
not clear if an icon was leading to more specific information, to sign up for updates, or to
ask questions. Further studies are needed to examine the relationships between
searchability and interactivity, as well as to explore the reasons for negative consumer
attitudes towards interactivity in food safety communication.
Linkability, despite being experienced differently by consumers from searchability in
research by Ma et al. (2016a), was shown to be somewhat indistinguishable from
searchability as a construct as it is highly correlated with searchability in EFA and its
contribution in explaining variations beyond what’s been accounted for by searchability
is minimal in SEM. Linkability is similar to searchability, so it does not significantly
impact behavioral intention beyond the effect of searchability; but this does not explain
why consumers experience them differently. For example, in research by Ma et al.
(2016a), consumers indicated that the use of links helps them to find additional
information (e.g. explanations of a concept), which is different from locating the needed
information (searchability). It is possible that linkability has multiple dimensions that are
more closely aligned with other characteristics of the websites such as searchability,
interactivity, and/or information quality. Further explorations of consumer experience are
needed. Another possibility for the insignificant influence linkability has on behavioral
intention is that consumers do not prefer to see linkability features on a website as these
features can be perceived as slowing down the information seeking process and making it
more difficult. For example, when linked to an outside sources, consumers have to
reevaluate the information quality, and this requires additional efforts. It is also possible
that the linked information is not as salient or is hard to understand. For instance, if
141
CDC’s website is linked, it is generally hard to find localized and specific information.
And sometimes, despite the good quality, information can be difficult to comprehend on
these websites as technical terms are often used to ensure accuracy.
With these in mind, when designing and using the website to communicate about
foodborne illness outbreaks, it may be best to create a direct link to the “what’s going on
now” page where information of the ongoing outbreaks are updated (so consumers do not
need to search extensively for outbreak updates). It may also be good to create an
“alerting box” on the home page – informing consumers of ongoing outbreaks, if there
are any or reassuring consumers that there are currently no ongoing outbreaks. Food
safety communicating in this sense is similar to marketing communication; if you want
consumers to use the information, it is important to make it easy for them to find the
information – and one way to achieve this is to draw consumer attention to the most
important and relevant information. Thus, using redundant features to highlight the most
relevant information can be very helpful in improving perceived process efficacy for
consumers. This is not to suggest putting all details on the home page or even the “what’s
going on now” update page. Rather, keep the updates simple and easy to ready so
consumers can quickly determine what to do to protect themselves and links can be used
to direct consumers to more detailed information if they are interested.
In examining EPPM’s utility, it was found that when confronted when a threat, perceived
self- and response efficacy are predictive of information usage intention. This finding
supplies support for EPPM’s power in understanding consumer food safety
informationseek behaviors. Theoretically, this study proposed and tested two additional
efficacy components beyond what has previously been included in EPPM. Process and
message efficacies proved to be important in determining consumers’ behavioral
142
intentions, particularly in online communication. This finding contributes to an expanded
understanding of efficacy beyond what was offered in the EPPM, and it opens doors for
further use of EPPM in computer-mediated communication.
This study proposed different scenarios that varied in terms of website characteristics.
The manipulations were tested to be successful, and the design of the scenarios is flexible
in that additional website features can be easily incorporated. Thus this research design,
particularly the setup, may be used in future studies pertaining to food safety
communication online or user experience studies more broadly.
5.7 Limitations and future studies
Limitations of this study include that no observations were made of users’ actual
experiences using the website. All the reactions were prompted using scenarios. This is a
limitation in terms of realism of this study, but it also means all other potential
confounding variables were controlled. Second, the present study only selected certain
website characteristics that emerged out of previous studies as important (Ma et al.,
2016a) to test in the scenarios. Though the list may not be comprehensive, this setting
provides an example for future exploration to incorporate additional website
characteristics. Additionally, new services may emerge that would improve consumers’
usage experience. Thus practitioners are encouraged to continuously monitor for new
features that may impact consumers’ perceptions and evaluations of the websites. Third,
the scenarios were developed to solicit two possible levels (high vs. low) of perceptions.
Because of this, while showing situational details, the wording could have been
suggestive, and in turn, influenced participants to respond with more extreme evaluations.
143
However, the use of scenarios in this study successfully probed significantly different
levels of perceived searchability, linkability, interactivity, trustworthiness, timeliness, and
saliency and thus allowed the investigation of how such perceptions influence efficacy
perceptions and later behavioral intentions. Fourth, instead of actual behaviors, this study
examined behavioral intentions. While a good predictor of behaviors (Ooms, Jansen, &
Hoeks, 2015), intentions can be different from actual behaviors (Fishbein & Ajzen,
1977). Fifth, this study only considered situations where consumers were confronted with
a threat that they perceived to be high. This, as discussed in the literature review, was
done to focus attention on efficacy and media effect. In the future, studies comparing
consumer experiences when confronted with a high vs. a low level of threat may also be
valuable.
Recommendations for future studies include qualitative observations of consumers’ actual
behaviors that would be helpful in understanding the reasons behind the relationships
shown in this study. Second, in this study, searchability was the only factor influencing
process efficacy; more studies are needed to verify and expand our understanding of other
factors that may influence process efficacy. Third, scholars are encouraged to investigate
the best update interval in communicating foodborne illness outbreak information and
further explore the relationships between information timeliness and message efficacy,
and later behavioral intention or behaviors. Fourth, with the advancement of technology,
it is expected that new features or services will emerge that can significantly impact
users’ information-seeking experience online; therefore it is necessary to continuously
include and test the impacts of new features. It is also important to keep in mind that with
new features, new perceptions can emerge. Fifth, this study focused on consumer
experience using websites because websites are consumers’ preferred platform. Social
144
media, despite its potential for food safety communication, is not the preferred platform
for consumers at this time (Ma et al., 2016b). Websites are also the main online
communication platform for the majority of the food safety information. As social media
presents promising benefits in enhancing food safety communication, future studies may
explore similar research questions using social media as the platform.
It will also be valuable to compare results with social media as the platform to the results
of this study. The results can shed light on relative advantages of each platform and supply
information for setting up strategies to integrate the use of both platforms. Sixth, setting up
an “ideal” website and collecting user reactions to compare with user experience with
existing websites can provide additional valuable insights. Seventh, this study examines
six different perceptions towards website characteristics in one setting; future studies may
isolate one perception at a time so the possible interactions among the perceptions are
controlled. Eighth, examining the potential moderating roles of social and demographic
characteristics (e.g. gender, age, education) can be valuable especially it is found that food
safety knowledge can have a impact on efficacy perceptions. Ninth, perceptions of threat
and their roles in food safety communication need to be further explored. More specifically,
what is perceived as a threat to consumers, what factors influence threat perception (e.g.
age, gender, and past experience), and whether high vs.
low threat perceptions lead to different behavioral intentions all need more attention.
5.8 Conclusion
This study explored media’s effect on food safety communication. Specifically, this study
examined how website characteristics, through efficacy perceptions, influence
consumers’ intention to use the communicated information. The results support that not
only the information quality, but also how the information is communicated can have a
145
significant impact on consumers’ behavioral intentions. Website features that directly
related to searchability (e.g. search box and site map) and saliency (provide specific
information about foods and locations involved) are perceived to be most influential and
should be considered in website design. The importance of communicating high quality
information over an easy-to-use media seems intuitive, yet the question of how to best
accomplish this task is very hard to answer. This study provides insights and hopefully
will help guide the efforts of practitioners and researchers alike.
