WHAT’S IN AN APOLOGY? VIRALITY AND SOURCE EFFECTS ON EVALUATIONS OF
CRISIS RESPONSE STRATEGIES VIA FACEBOOK
By
Shupei Yuan

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Advertising – Master of Arts
2013

ABSTRACT
WHAT’S IN AN APOLOGY? VIRALITY AND SOURCE EFFECTS ON EVALUATIONS OF
CRISIS RESPONSE STRATEGIES VIA FACEBOOK
By
Shupei Yuan
Social networking sites (SNSs) have become important tools for organizations and companies to
communicate with consumers and constituents during crisis situations. The current study
investigated the persuasive effects of message virality and source type on evaluations of apology
crisis responses posted on Facebook. The study employed a 2 (source type) x 2 (number of likes)
x 3 (repetition) mixed factorial experiment. Participants were exposed to three crisis scenarios,
each followed by an apology crisis responses posted on Facebook. Participants were exposed to
the crisis response posted either on the CEO Facebook page or the company’s official Facebook
page. Additionally, participants either saw the status update accompanied by a low or high
number of likes. The study’s findings were discussed in relation to crisis response strategies and
persuasion models. Practically, the study’s result that participants’ evaluation of perceived source
credibility varies with the interaction between source type and virality, trustworthiness has been
showed the influence on subjects’ attitude. The results showed the insignificant difference of the
effects of source type or virality on attitude or intention. It also showed the significant
relationship between perceived source credibility and attitude.

ACKNOWLEDGMENTS

First, a great thanks to my thesis advisor Dr. Saleem Alhabash, without his help with
generating this study idea, designing the structure of literature review, collecting data and
analyzing it, this study couldn’t have been done. His advice guided me from the beginning and
taught me how exciting research can be. Moreover, he has always been very supportive during
my first research experience, which is extremely important for me. Thank you very much!
I also would like to thank my two committee members for their great efforts in my study,
Dr. Hairong Li and Dr. Elizabeth Taylor Quilliam. Dr. Li and Dr. Quilliam offered great feedback
to my proposal, and both of them helped me to diligently modify my paper. Dr. Li suggested a lot
on my conceptual framework building, and Dr. Quilliam helped me to make this study more
rigorous. I have enjoyed working with them very much!
Another special thanks to Dr. Quilliam and Sooyong Kim for helping me collect data by offering
students extra credit in their classes. Without their help, I could not receive this big database.

iii

TABLE OF CONTENTS

LIST OF TABLES........................................................................................................................ vi
LIST OF FIGURES ................................................................................................................... viii
INTRODUCTION......................................................................................................................... 1
LITERATURE REVIEW ............................................................................................................. 2
Crisis Management and Communication ............................................................................... 2
Apology: ‘Sorry seems to be the Hardest Word’ ................................................................... 7
Social media: A PR Platform ................................................................................................... 9
SNSs: Potential Venue for Crisis Communication ............................................................... 10
Theoretical Framework: Persuasion Theories ..................................................................... 13
Sources of Crisis Communication ......................................................................................... 16
Virality as a Message Feature ................................................................................................ 20
METHOD .................................................................................................................................... 24
Design and Participants .......................................................................................................... 24
Independent Variables ............................................................................................................ 24
Source Type.......................................................................................................................... 24
Number of Likes .................................................................................................................. 24
Repetition ............................................................................................................................. 25
Dependent variable ................................................................................................................. 25
Attitude toward Crisis Response Message ........................................................................ 25
Attitude toward Company in Crisis .................................................................................. 25
Viral Behavioral Intentions (VBI) ..................................................................................... 26
Perceived source credibility-attractiveness....................................................................... 26
Perceived source credibility-trustworthiness ................................................................... 27
Perceived source credibility-expertise ............................................................................... 28
Stimuli and Pretest .................................................................................................................. 28
Main Study Procedure ........................................................................................................ 30
Manipulation Check ........................................................................................................... 31
Data Reduction and Analysis ................................................................................................. 32
RESULTS..................................................................................................................................... 33
Descriptive Results .................................................................................................................. 33
Attitude toward Status Update .............................................................................................. 33
Attitude toward Company in Crisis ...................................................................................... 34
Viral Behavior Intentions ....................................................................................................... 35
Predicting Attitudes and VBI from Perceived Source Credibility...................................... 36
Perceived source credibility and Attitude toward Status Update ....................................... 37
Perceived source credibility and attitude toward Company in Crisis ............................... 40
Perceived source credibility and Viral Behavior Intention ................................................. 43
iv

DISCUSSION .............................................................................................................................. 47
THEORTICAL AND PRACTIAL IMPLICATION ................................................................ 49
LIMITATION AND FUTURE STUDY .................................................................................... 50
CONCLUSION ........................................................................................................................... 51
APPENDICES ..................................................................................Error! Bookmark not defined.
Appendix 1. Facebook status update stimuli ........................................................................ 53
Appendix 2. Pre-test Questionnaire ...................................................................................... 59
Appendix 3. Main test questionnaire .................................................................................... 71
FOOTNOTE ................................................................................................................................ 79
REFERENCE .............................................................................................................................. 81

v

LIST OF TABLES

Table 1. Crisis types by crisis cluster according to SCCT .............................................................. 5
Table 2. Factor analysis and reliability for attitude toward crisis response message. ................. 25
Table 3. Factor analysis and reliability for attitude toward company in crisis. ........................... 26
Table 4. Factor analysis and reliability for attitude toward viral behavior intention. ................. 26
Table 5. Factor analysis and reliability for attractiveness. .......................................................... 27
Table 6. Factor analysis and reliability for trustworthiness. ........................................................ 27
Table 7. Factor analysis and reliability for expertise. .................................................................. 28
Table 8. Descriptive statistics of pre-selected crisis cases. .......................................................... 29
Table 9. Descriptive statistics of pre-test selected CEO images and company logos. .................. 30
Table 10. Manipulation check on number of likes. ....................................................................... 32
Table 11. Tests of between-subject effects on attitude towards status update. .............................. 34
Table 12. Mean for attitudes towards status update. .................................................................... 34
Table 13. Tests of between-subject effects on attitude towards company in crisis....................... 35
Table 14. Mean for attitudes towards company in crisis. ............................................................. 35
Table 15. Tests of between-subject effects on viral behavior intention. ....................................... 36
Table 16. Mean results for viral behavior intention. ..................................................................... 36
Table 17. Summary of linear regression analysis for attitude towards status update (overall data).
.............................................................................................................................................. 37
Table 18. Summary of linear regression analysis for attitude towards status update (spilt file by
“source”) .............................................................................................................................. 38
Table 19. Summary of linear regression analysis for attitude towards status update (spilt file by
“likes”) ................................................................................................................................. 39
Table 20. Summary of linear regression analysis for attitudes toward status update (spilt file by
“likes” and “source”) .......................................................................................................... 40

vi

Table 21. Summary of linear regression analysis for attitude towards company in crisis (overall
data) ...................................................................................................................................... 41
Table 22. Summary of linear regression analysis for attitude towards company in crisis (spilt file
by “source”) ......................................................................................................................... 41
Table 23. Summary of linear regression analysis for attitude towards company in crisis (spilt file
by “likes”)............................................................................................................................. 42
Table 24. Summary of linear regression analysis for attitudes toward company in crisis (spilt file
by “likes” and “source”) ...................................................................................................... 43
Table 25. Summary of linear regression analysis for viral behavior intention (overall data) ..... 44
Table 26. Summary of linear regression analysis for viral behavior intention (spilt file by
“source”) .............................................................................................................................. 44
Table 27. Summary of linear regression analysis for viral behavior intention (spilt file by “likes”)
.............................................................................................................................................. 45
Table 28. Summary of linear regression analysis for viral behavior intention (spilt file by “likes”
and “source”) ....................................................................................................................... 46

vii

LIST OF FIGURES

Figure 1. CEO response with high likes (case 1). ......................................................................... 53
Figure 2. CEO responses with high likes (case 2) ........................................................................ 53
Figure 3. CEO response with high likes (case 3) .......................................................................... 54
Figure 4. CEO response with low likes (case 1) ........................................................................... 54
Figure 5. CEO response with low likes (case 2) ........................................................................... 55
Figure 6. CEO response with low likes (case 3) ........................................................................... 55
Figure 7. Organization response with high likes (case 1) ............................................................. 56
Figure 8. Organization response with high likes (case 2) ............................................................. 56
Figure 9. Organization response with high likes (case 3) ............................................................. 57
Figure 10. Organization response with low likes (case 1) ............................................................ 57
Figure 11. Organization response with low likes (case 2) ............................................................ 58
Figure 12. Organization response with low likes (case 3) ............................................................ 58
Figure 13. CEO Images used in study .......................................................................................... 67
Figure 14. Company Logos........................................................................................................... 69
Figure 15. Facebook Status Test Material Example ..................................................................... 75

viii

INTRODUCTION
In 2012, about 66% of online users reported using social networking sites (SNSs), with
Facebook on top with over a billion users and most traffic (Dugan, 2012). The popularity of
SNSs, wide adoption by individuals, companies, and organizations has been mirrored by
scholarly attention attempting to understand why, how, and with what effects individuals use
SNSs, especially within the context of marketing, advertising, and public relations (Austin, et al.,
2012; Dekay, 2012; Nelson-Field, Riebe & Sharp 2012; Vejacka, 2012; Yousif, 2012).
Commercial companies and non-profit organizations have been leveraging the popularity,
power, and multi-modality affordances (i.e., text, audio, and video) of SNSs to deliver
information, build online communities, and respond to crises (Dekay, 2012). Additionally,
around 70% of respondents expressed that they expect organizations to use social media websites
and services to update crisis information and respond to crises. The use of social media has been
a key tactic in responding to recent crises like the 2010 British Petroleum (BP) Gulf of Mexico
oil spill, Blackberry Network Outage in 2011, Sony PlayStation Network attack in 2011, and
Amazon web service outage in 2011 (Hendrick, 2011).
Crisis managers are realizing the influence of social media. While we know about such
practices, an understanding of the effectiveness of using social media to disseminate crisis
responses is still understudied. To this end, the current study investigates the effects of crisis
communication within the context of SNSs. More specifically, the study focuses on the effects of
source type and message virality on evaluations of crisis response strategies. The experiment
systematically varied the source (CEO or company page) and virality (number of likes) in
apology crisis responses shared on Facebook for a fictitious organization.

1

LITERATURE REVIEW
This section reviews past research and theoretical frameworks in relation to crisis
management and crisis communication. Second, the section conceptualizes apology as a
prominent crisis response strategy. Third, the study provides a review of major trends in the use
of social media and social networking sites (SNSs) in the practice of public relations. Fourth,
reviewing major persuasion models provides the theoretical framework of the study. Fifth, the
independent variables – virality and source type – are conceptualized, and the study’s hypotheses
are proposed.
Crisis Management and Communication
Crisis management has become a popular topic of inquiry. Researchers used various
methods –from case studies to experimental studies – to classify crises, summarize crisis
response types, develop crisis management strategies, and investigate the effectiveness of postcrisis management strategies (Coombs, 2007a; Lalonde, 2007; Grunig, 1992; Benoit & Drew,
1997). These studies have helped in building several theoretical frameworks for crisis
communication that further guide companies or organizations in overcoming crisis situations.
The following section provides a conceptual definition of a crisis and reviews different
theoretical approaches to crisis response strategies.
Crises are defined as “major, unpredictable and unexpected events that threaten to harm
the organization or its stakeholders” (Coombs, 1999, p. 3). Lalonde (2007) described four aspects
that are important to consider when looking at a crisis situation: (1) its origin (whether it is a
natural or human-made crisis); (2) consequences (the result brought by the crisis); (3) underlying
cause (the reason causing the crisis); and (4) level of risk (the severity of damage). Post-crisis
communication (i.e., crisis response) helps organizations and companies to avoid or diminish the

2

reputational damage resulting from a crisis (Coombs & Holladay, 2005).
Crisis communication and management are prominent public relations functions. Defined
as the “management of communication between an organization and its publics”(Grunig & Hunt,
1984, p. 6), public relations deals with managing publics’ opinions and perceptions about an
organization, which ultimately influence an organization’s future. While a nascent field of
inquiry, public relations theory can be grouped into two main approaches: the excellence theory
and the contingency theory. Grunig’s (1992) excellence theory (ET) conceptualizes public
relations as a strategic management function by which organizations manage internal and
external communication. Its major premise lies in making public relations more effective for
practitioners (Grunig, 1992). The excellence theory proposes four different public relation
models that vary in communication symmetry and the entities involved in communication. First,
the press/agentry/publicity model uses persuasion or manipulation to influence the audience to
behave in the way the organization desires. Second, the public information model uses press
releases or other one-way techniques to deliver information. Third, the two-way asymmetrical
model uses persuasion and manipulation and disregards public’s feeling. Finally, the fourth
model, the two-way symmetrical communication model, focuses on negotiating and solving
problems between organization and publics, and is deemed the best – or excellent – model for
managing public relations efforts.
While the excellence theory favors the two-way symmetrical model of communication,
the contingency theory argues otherwise (Cancel, Cameron, Sallot, &Mitrook, 1997). Cancel and
colleagues (1997) argued that contingent factors affect organizations’ strategies in dealing with
publics. Contrary to excellence theory, the contingency theory posits that true excellence should
instead facilitate public relations to pick the most appropriate strategies (Cancel et al., 1997).

3

These strategies range on a continuum from complete accommodation to complete advocacy,
where the contingent factors affect the placement of an organization’s stance on the
accommodation-advocacy continuum. Accommodation refers to an organization’s stance that
yields to the needs and demands of the stakeholders, much like the two-way symmetrical model
of the excellence theory, whereas advocacy stances are ones that primarily reflect an
organization’s position without any regard to stakeholders.
The difference between the two theoretical approaches does not lie in having different
foci, but rather in conceptualizing what constitutes as best practices in communicating with the
public. The contingency theory argues that based on situational and predispositional elements
specific to the organization’s internal and external factors, taking an advocacy stance could
exemplify excellent communication. Both approaches are extremely relevant to managing and
responding to crises as they emphasize that there cannot be a single way to communicate with
publics. Responding to crises requires adequate understanding of the situation, the organization,
and the environment. Both approaches explain the complex and dynamic process of managing
relationships, crises, reputations, and competition at the organizational level. The current study
primarily deals with crisis communication. Coombs’ (2007a) situational crisis communication
theory (SCCT) and Benoit’s (1997) image repair theory are reviewed below.
Coombs (1995) argued that the process of crisis-response selection involves identification
of the crisis type, gathering additional evidence, understanding the severity level, reviewing
performance history (whether this issue happened many times in this field), and ends with an
appropriate response to the crisis situation. Coombs (2007a) proposed SCCT as a framework to
understand the nature of crises and various types of crisis responses. According to SCCT, crises
are divided into three major types based on attributions of crisis responsibility: victim, accidental,

4

and intentional clusters. Crises in the victim cluster include crises where organizations are also
victims of the crisis. Examples of victim cluster crises include natural disasters and product
rumors that are outside the company’s control. Second, accidental cluster crises are those where
companies or organizations have minimal contribution to the crisis, such as technical-error
accidents or product harm instances. Third, preventable cluster crises are when organizations
“knowingly place people at risk, took inappropriate actions or violated a law/regulation”
(Coombs, 2007a, p.168) and “where companies have strong attribution to the crisis responsibility”
(Coombs, 2007b, p.137). Examples of intentional crises would be human-error accidents or
product harm crises (see Table 1).
Table 1.
Crisis types by crisis cluster according to SCCT
Victim Cluster: an organization is also a victim of the crisis.
(Weak attribution of crisis responsibility = Mild reputational threat)
Natural disaster: Acts of nature damage an organization such as an earthquake.
Rumor: False and damaging information about an organization is being circulated
Workplace Violence: Current or former employee attacks the current employees onsite.
Product Tampering/Malevolence: External agent causes damage to an organization.
Accidental Cluster: organizational actions leading to the crisis were unintentional
(Minimal attribution of crisis responsibility)
Challenges: Stakeholders claim an organization is operating in an inappropriate manner.
Technical-error accidents: A technology or equipment failure causes an industrial accident.
Technical-error product harm: A technology or equipment failure causes a product to be
recalled.
Preventable Cluster: organization knowingly placed people at risk, took inappropriate actions or
violated a law/regulation
(Strong attribution of crisis responsibility)
Human-error accidents: Human error causes an industrial accident.
Human-error product harm: Human error causes a product to be recalled.
Organizational misdeed with no injuries: Stakeholders are deceived without injury.
Organizational misdeed management misconduct: Laws or regulations are violated by
management.
Organizational misdeed with injuries: Stakeholders are placed at risk by management and
injuries occur.
Source: Coombs (2007a, p.168, 2007b, p.137)
Based on the different crisis types, Coombs (2007a) suggested three major response
5

strategies that match crisis types respectively: denial, diminishing, and rebuilding strategies.
Denial is when companies or organizations “attempt to remove connections between them and
the crisis” (p. 171). In such cases, the crisis manager asserts that there is no crisis happening by
attacking the accuser, simply denying the crisis, or scapegoating. In using diminish strategies,
managers minimize the severity of a crisis and claim lack of control over crisis. They let people
view the crisis less negatively and claim the severity of crisis is reduced. Examples of diminish
strategies include finding excuse and justification. Finally, rebuild strategies “offer material
and/or symbolic forms of aid to victims” (Coombs, 2007a, p.172). Crisis managers declaim to
take full responsibility of the crisis by using rebuild strategy, such as compensation and apology.
Coombs’s framework for understanding crisis response strategies rests upon Benoit’s
(1997) earlier conceptualization regarding the process of image repair. In image repair theory,
Benoit argued that restoring a company/organization’s image or reputation that has been
hampered is an important organizational function. Image repair strategies are formed into a
comprehensive model with five response options: denial (refuse to commit the act), evasion of
responsibility (reduce the responsibility of act), minimization (reduce offensiveness of act),
corrective action (offer to repair damages), and mortification (ask for forgiveness). Benoit
explained that the key of image repair strategies is to “consider the nature of attacks or
complaints that prompt such responses or instigate a corporate crisis” (Benoit, 1997, p. 178),
which means strategies should match with situations.
Much of the research on crisis communication takes a case study, descriptive approach
(Maenaka, 2007; Klikauer, 2012; Barker, 2011). The dynamic nature of crises, as well as the
complex structure of organizations, makes it hard to predict which crisis response type is most
effective in alleviating the reputation damage resulting from a crisis. Critical scholars concluded

6

that an effective response helps the company or organization diminish the negative effect of the
crisis by salvaging the brand image and stock price, among other assets (Benoit & Drew, 1997;
Coombs &Holloway, 2005). In cases where the organization is guilty, Kim, Avery and Lariscy
(2009) argue that the ultimate resort is to apply rebuild strategies as the most effective way to
rebuild the company/organization’s reputation and image. For example, musician Dave Carroll
wrote and performed a song titled ‘United Breaks Guitars’ and posted it on YouTube after his
guitar was broken during a trip on United Airlines in 2008 (FOX news, 2011; CBC news, 2009).
United Airlines initially refused to take responsibility because Carroll failed to report the incident
within 24 hours (Cosh, n.d). After the YouTube video went viral, United Airlines issued a public
statement of apology and engaged in corrective action, however, responding in a slow manner
negatively affected the company’s image. Some argue that if United Airlines had responded with
an apology earlier, the crisis would have been avoided (Greenfield, 2009). Today, when online
content can quickly get viral, companies also try to promptly respond to potential crises to
minimize harm to a company’s reputation.
Apology: ‘Sorry seems to be the Hardest Word’
Apology is probably the most advocated crisis response strategy (Patel & Reinsch, 2003;
Kim, Avery, &Lariscy, 2009; Claey, Cauberghe, & Vyncke, 2010; Bradford & Garrett, 1995,
Dean, 2004). It is classified as a mortification strategy in Benoit’s typology (1997), and as a
rebuild strategy in Coombs’ SCCT (2007a). Apology refers to situations when the organization
confesses wrongdoing, accepts responsibility for the crisis, and begs for forgiveness (Benoit &
Drew, 1997; Fuchs-Burnett, 2002). An apology includes expressions of remorse,
acknowledgement of responsibility, offers of reparation, forgiveness request, and a promise of
not repeating the offense (Buttny, 1993; Benoit & Drew, 1997; Patel & Reinsch, 2003;Fuchs-

7

Burnett, 2002). An effective apology acknowledges “that a moral norm or an understanding of a
relationship was violated, explanation of why the offense was committed, and a communication
that the actions were not intentional” (Lazare, 1995, p. 40).
Researchers have shown that apology is prevalent and has positive effects as a crisis
response strategy. Patel and Reinsch (2003) analyzed crises between 1990 and 2000, and found
that apology was the most common strategy. They also found that apology helped in “shaping
corporate reputation, facilitating forgiveness or private settlement, evidence for the plaintiff
(admission of guilt), and evidence for the accused (reduction of penalty)” (Patel & Reinsch, 2003,
p. 16-19). Kim, Avery and Lariscy (2009) reached similar conclusions from their content analysis
of crises between 1991 and 2009, where apology was evaluated as the most effective crisis
responses strategy compared to others. They further suggested that even though, theoretically,
strategies should match with situations, companies or organizations should review a crisis more
comprehensively, rather than just applying a “best practice” approach. This is where
experimentation is most useful. A number of studies have demonstrated that apology has the
strongest positive effect on post-crisis perceptions of the organization (e.g., Bradford &Garrett,
1995; Dean, 2004). Another study about the impact of different crisis response strategies found
that respondents have more positive attitudes toward apology responses compared to other
strategies such as denial or no response (Claey, Cauberghe, & Vyncke, 2010).
Despite differences in crisis situations, there seems to be an agreement that apology
might be the most effective response in a crisis. Past theorizing on public relations as an
organizational function to manage crises and competition (Cancel, et al., 1997, Benoit, 1997;
Coombs & Hollodays, 1996) argued that organizations move their stances according to the
interplay of multiple situational and predispositional factors. It is plausible that Gruing’s (1992)

8

excellence theory might hold up the test of time in the sense that apology is favored, if not
expected, from organizations. Based on the common usage and the effectiveness of apology
strategy, the current study focused on apology strategy in designing the experimental stimuli.
To summarize, scholars have discussed crisis communication for decades. It has been
developed into different categories based on the variety of crisis situations and media
environment. It offers researchers the basis for further studies when facing complex issue or
environment. Especially in the social media era, post-crisis communication has to face
unexpected opportunities and challenges posed by social media. The following section focuses
on the use of social media services and SNSs as marketing, advertising, and public relations tools.
Social media: A PR Platform
Social media have changed the relationship between companies and their customers.
Managing relationships have changed in an environment of fast-paced, readily available
information. The change in audience structure and information receiving behaviors bring
opportunities for organizations or companies to communicate with the public.
Social media are defined as “a group of Internet-based applications, which allow the
creation and exchange of user generated content” (Kaplan & Haenlein, 2010, p. 67). The use of
social media for business, marketing, and public relations purposes has garnered increasing
attention by researchers. Social media help companies and organizations create and strengthen
relationships, connect people with similar interests, and build brand awareness (Knobel,
Lankshear, 2008). Barnes and Mattson (2008) also found that it takes less time for social media
to reach out to potential or existing customers compared to traditional media such as television or
radio. With less space and time limitations, social media deliver increasingly precise information
faster to target audiences. Naveed (2012) conducted a study in Pakistan to examine the effects of

9

social media on consumers’ buying behavior and found a positive relationship between
organizational communication and users on blogs. The author found that 73% of respondents
agreed that blogs have been changing the way organizations communicate with customers. Social
media are becoming essential platforms where companies and organizations practice public
relations.
Different from research on the effects of mediated communication that focus primarily on
message content rather than message channels, studies using social media amplify the
importance of information form, particularly in a crisis situation (Austin, Liu, & Jin, 2012).
Austin and colleagues (2012) applied the social mediated crisis communication model that
focuses on the interaction between social media creators, followers and inactives. Creators are
social media users who create information, followers are those who consume information, and
inactives consume information indirectly through word-of-mouth channels from social media
followers. Austin and colleagues (2012) found that participants mainly used social media during
crises to get “insider information” as a practice of self-knowledge seeking and communicate with
their family and friends to make sure they are safe and alright (p. 197). They also found that
participants seek information through friends’ Facebook pages and follow updates on Facebook
fan pages more than traditional media during crises. While intriguing, these findings call for
further investigation of factors that affect audiences’ evaluations of crisis response strategies and
the effects of information source and form in relation to yielding to arguments made in crisis
responses.
SNSs: Potential Venue for Crisis Communication
Scholarship on social media has matured over the past few years. This scholarship
maturation was matched by increased sophistication among industry professionals in

10

understanding the specificities of different genres of social media. Sterne (2010) classified social
media sites and services into six categories: (1) social network sites (SNSs) like Facebook; (2)
microblogging sites like Twitter; (3) media sharing sites like Flickr and YouTube; (4) blogs like
WordPress; (5) forums and message boards like Craigslist; and, (6) review and opinion sites like
eBay. The current study mainly focuses on SNSs as the context of delivering apology responses
to a crisis situation.
boyd and Ellison (2007) defined SNSs as “web-based services that allow individuals to (1)
construct a public or semi-public profile within a bounded system, (2) articulate a list of other
users with whom they share a connection, and (3) view and traverse their list of connections and
those made by others within the system”(p. 211). SNSs constitute a type of computer-mediated
communication (CMC) technology that provides an online forum for people who share similar
interests to exchange information, such as pictures, messages, or hyperlinks (Sledgianowski &
Kulviwat, 2009). Facebook is the most popular SNS and social media site with one billion users,
compared to Twitter’s 550 million, Google plus’s one million users, and LinkedIn’s 150 million
users (Dugan, 2012). Facebook receives 7.01 billion visitors every month (Dugan, 2012).
According to the Pew Internet and American Life Project, about two-thirds of online adults and
three-quarters of young adults aged 18 to 29 use SNSs (Lenhart, Purcell, Smith, & Zickuhr,
2010). Brenner (2013) recently reported that an extreme majority of young adults (92%) reported
using SNSs, a percentage that grew from 9% in 2005. The majority of these SNS users (89%)
report having a Facebook account.
A distinguishing characteristic of SNSs is the use of SNSs for hedonic purposes such as
enjoyment and ease, rather than utilitarian purposes that focus on specific usage intentions
(Sledgianowski & Kulviwat, 2009). Sledgianowski and Kulviwat (2009) expanded six aspects of

11

SNS usage intention: playfulness, development of critical mass, increased trust, yielding to
normative pressures, perceived ease of use and perceived usefulness. These factors might be
critical in attempting to use Facebook for marketing, advertising, and public relations purposes,
as practitioners need to understand not only how users use SNSs, but also, why they are using
them.
A number of scholars argued that Facebook, among other SNSs, has four potential
benefits to marketers, advertisers, and public relations practitioners (Austin, et al., 2012; Dekay,
2012; Nelson-Field, Riebe & Sharp 2012; Vejacka, 2012; Yousif, 2012). First, it could help
companies and organizations increase brand awareness, thus leading to potential higher sales and
market share. Second, Facebook is used as a direct advertising platform. Third, Facebook could
be used as an avenue to build and maintain fan-based platforms. Finally, and possibly most
importantly, Facebook is used as a platform for companies and organizations to engage in direct
two-way communication with existing or potential customers.
Nelson-Field and colleagues (2012) explored customer behaviors of Facebook users who
are “fans” of brands. They found that the majority of a brand’s fans on Facebook are also heavy
buyers, thus suggesting that a Facebook fan base provides a good opportunity to listen and
communicate with existing buyers that leverages a brand’s social network and increases the
number of brand advocates. Within the context of advertising, Yousif (2012) showed that
participants thought advertising on Facebook not only was considered an effective promotional
tactic, but it was also associated with the provision of trustworthy and reliable messages.
Additionally, participants considered Facebook advertising appealing and exciting.
With 79% of Fortune 100 companies using social media to communicate with their
customers or stakeholders, Dekay (2012) argued that the affordances of Facebook, specifically

12

the comments section, allows consumers to communicate directly with companies and brands.
This open-channel communication, however, is coupled with companies’ willingness to censor or
ignore critical feedback instead of responding to negative comments (Dekay, 2012).
To summarize, most studies on business-to-consumer communication social media and SNSs
focuses on information owners’ perspective, such as analyzing content on corporations’ website
or asking industry professionals about how they use social media in their work. Contrary to such
approaches, the current study focuses on message receivers. More specifically, the current study
is set to investigate the effects of virality (low vs. high likes) and message sources (company vs.
CEO) on participants’ attitudes and behavioral intentions in relation to a hypothetical crisis
situation. Ultimately, a crisis response aims at changing people’s attitudes in relation to a crisis
situation. To address this, the following section conceptualizes the effects of crisis responses
using different persuasion models.
Theoretical Framework: Persuasion Theories
The effectiveness of a crisis response strategy is reflected by the formation of and change
in audiences’ attitudes toward the company/organization following a crisis. In the current media
environment, audience attitudes and behaviors are shaped by many factors such as the motivation
to attend to a message, previous knowledge about the company/organization, and message
characteristics. All these factors, according to different persuasion models, affect whether
individuals accept the message’s arguments, and in turn change their attitudes and behaviors
toward the object of persuasion. The current study argues that a crisis response is a form of
persuasive message that an organization publicizes during and/or after a crisis, where the
outcome of responding to a crisis is attitudinal and behavioral change. Additionally, the study
posits that two features of a crisis response shared on a popular SNS: the source of an apology

13

and the message’s virality, are bound to affect individuals’ attitudes and behavioral intentions in
relation to the company/organization in crisis.
Most commonly, attitude is defined as an evaluative response that involves a knowledge
function, such as beliefs, feelings, or past behavior (Fazio, 1990; Zanna & Rempel, 1988).
Behavior intentions are indications of an individual’s readiness and willingness to perform a
given behavior and attain a goal (Ajzen, 1991).
The study of persuasion has been closely tied to theories of information processing like
the elaboration likelihood model (ELM; Petty & Cacioppo, 1983; Petty, Brinõl & Priester, 2008;
Petty & Wegener, 1999) and the heuristic-systematic model (HSM; Chaiken, 1980; Eagly &
Chaiken, 1993). ELM assumes that certain message features, as well as the receiver’s motivation
and ability to process information, activate either the peripheral or central route, thus affecting
the receiver’s acceptance and yielding to a persuasive messages (Petty & Cacioppo, 1986). In
peripheral route processing, individuals use simple heuristic cues or informational indicators,
such as a model’s attractiveness or the source’s perceived credibility, to assess the believability
of a message. Central route processing involves greater level of deliberation of the message’s
persuasive arguments, where users have utilitarian cues based on previous knowledge and
experience and use those to evaluate the quality of the message’s arguments to reach the
persuasive outcome of attitude change. For example, Petty and Cacioppo (1990) reported an
interaction between issue relevance and the strength of persuasive argument in a message.
Individuals with higher issue-relevant cognitions were more receptive to stronger persuasive
arguments compared to those with little issue-relevant cognitions (Petty & Cacioppo, 1990).
Similarly, HSM divided types of information processing into heuristic and systematic
processing (Chaiken, 1980). In heuristic processing, individuals rely on cues associated with

14

memory or previous knowledge structures. On the other hand, systematic processing involves
comprehensive, cognitive analysis of the message and its arguments (Chaiken, 1980). Systematic
processing values perceived source credibility and message content regarding to persuasion
effect, Chaiken (1980) indicated that perceived source credibility affects the effect of persuasion
when readers are more under low-involvement, heuristic conditions, but not high, issue or
response-involvement conditions.
Models like ELM and HSM inferred that changes in attitudes are sufficient in affecting
an individual’s behavioral intention toward the object of persuasion. As introduced above,
behavioral intention is defined as individual’s readiness and willingness to perform a given
behavior and attain a goal (Ajzen, 1991). The theories of reasoned action (TRA; Ajzen &
Fishbein, 1980) and planned behavior (TPB, Ajzen, 1991) argued that the relationship between
attitudes and behaviors is not direct, and thus is mediated by an individual’s intention to perform
the behavior, or in other words, behavioral intentions. TRA combines attitude towards a given
behavior with expectations of others to predict intentions with a given behavior (Sheppard,
Hartwick & Warshaw, 1988), while TPB focuses on an individual’s perceived behavioral control,
defined as the perceived ability to perform a given behavior, as a precursor of performing the
behavior (Ajzen, 1985).
TPB argued that performing the desired behavior advocated in a persuasive message is
dependent on a number of factors: (1) attitude toward the behavior, which indicates people’s
willingness to perform the behavior; (2) the subjective norm, which indicates people’s
availability to perform the behavior; and (3) perceived behavior control, which refers to
resources such as time, money or skills that helps to succeed in doing this behavior (Ajzen, 1991).
As discussed above, attitude and behavioral intention are the major outcomes of

15

persuasion. Within the context of SNSs, persuasion takes a different form. Different from
traditional media, social media do not have “gatekeepers” to filter information before it is
disseminated to audiences, thus leading to information overload and uncertainty (Metzger,
Flanagin, &Medders, 2010). As a result, individuals can receive information from numerous
sources or channels. Additionally, the nature of attitude and behavioral change might take a
different direction. Instead of focusing on actual behaviors (e.g., purchase), online
communication allows for online behaviors ranging from liking, sharing, to commenting on a
message on a website like Facebook. The current study investigates the effects of source type
and message virality on attitudes toward the company, as well as offline and online behavioral
intentions related to that company. The following section focuses on the effect of sources within
the boundaries of persuasion.
Sources of Crisis Communication
Past research argued that individuals have higher levels of trust for messages from a
reliable source (Petty & Cacioppo, 1986, Dekay, 2012, Vejacka, 2012). With the growth in the
business uses of SNSs, companies and organizations have multiple accounts that serve similar
purposes. In addition to having an official page for a company, organization, or a brand, chief
executive officers (CEOs), other executives, and employees have professional presence on SNSs
like Facebook (Horton, 2009; Skeels, Grudin, 2009; Waters, Burnett, Lamm & Lucas, 2009). The
current study sheds the light on the potential effect of sources within the context of persuasive
messages shared on SNSs as a response to a crisis situation by exposing participants to crisis
responses posted by a human versus organizational source. Source, in general, is defined as the
entity to which information can be attributed. In other words, sources are originators of
information. In the current study, we define human source as one where the originator of the

16

information – in this case the crisis response – is an individual human being, where it is
manipulated as the CEO of the company in crisis. An organizational source refers to non-human
originators of information, where information is provided by an entity that resembles the
company in crisis.
Different types of sources have been showed different persuasive effects. A
conversational human source was found to have a significant effect on improving relationships
compared with an organizational source (Kelleher, 2009; Kelleher & Miller, 2006; Yang & Lim,
2009). It is prudent today to use conversational voice via social media when organizations try to
communicate with their consumers and stakeholders. In this situation, conversational human
source is described as “an engaging and natural style of organization and individuals in publics”
(Kelleher, 2009, p. 177). Park and Lee (2012) conducted a study to test participants’ perception
regarding human or organization presence within two organization types (nonprofit and for
profit), which used one nonprofit and one for-profit fictitious Twitter accounts. The authors
found that conversational human source was greater for organizations’ social network with a
human presence compared with organizational presence, which was compared by the human
profile pictures and name or organization logos and title in this study.
Source characteristics have been found to influence the effectiveness of persuasiveness
messages (Wilson & Sherrel, 1993; McGuire, 1985). Sources can be distinguished as “internal”
sources such as the communicator and messages or “external” sources such as medium channels
(Weiss, 1957). Therefore, the current study conceptualizes a source in relation to value of
information’s origin and its natural characters. Scholars have emphasized that different
dimensions of sources will influence the credibility of the communication message, such as the
origin of message, the message delivery channel and the endorser (Barnes, 1978; Schweiger,

17

2000; Sundar & Nass, 2001). Hovland, Janis, and Kelley (1953) have found that individuals will
express greater trust in information when it comes from a more reliable channel, or an originator
with authority or expertise. Therefore, source characteristics, such as content and originator, will
affect persuasion.
The study herein argues that varying the source type will result in varying degrees of
perceived source credibility, and thus different persuasive effects. Perceived source credibility
has been defined as “the extent to which a communicator is perceived to be a source of valid
assertion” (Hovland, Janis, & Kelley, 1953, p. 21). Past research has found that the credibility of
a corporation such as the message originator has both direct and indirect influence on audiences’
attitude toward the message and brand (Goldberg & Hartwick, 1990; Lafferty, Goldsmith &
Newell, 2002). Goldberg and Hartwick’s (1990) experiment found that the reputation (expertise
and trustworthiness) of a source had a significant effect on the subjects’ attitude change. Ohanian
(1990) added attractiveness into the scale to measure perceived source credibility. This factor is
significantly important when assessing the effects of celebrity endorsement, where attractiveness
plays an important role in people’s evaluation of the message source. Today, the attractiveness of
CEOs or founders of companies also play this role in a lot of companies today, famous company
leaders such as the co-founder of Apple Inc., Steve Jobs or founder of Facebook, Mark Zuckberg,
their personal symbols have become part of the company brands.
In the current communication environment, companies have multiple facets of their
online presence, facilitated by the affordances of social media. Companies and organizations are
not only having official, generic company pages on popular social media sites, but also, CEOs
are communicating directly to social media users on behalf of the company/organization. Park
and Lee (2012) conducted a study to test the effects of presence and organization type on

18

participants’ evaluations of the communication and organization. They exposed participants to
Twitter messages posted with a human or organizational source posted on Twitter pages of
nonprofit or for-profit organizations. They found that participants favored and felt more engaged
upon exposure to tweets written with a conversational human source as opposed to an
organizational source (Park & Lee, 2012). The current study argues that a CEO would be more
credible than a generic company page. Past studies showed that perceived source credibility to
positively affect persuasion (Bush, Moncriet, & Ziethami, 1987; Craig & McCann, 1978; Jin &
Yeo, 2011). Presumably, high source expertise is one of compelling cues to enhance persuasion,
whereas a weak compelling cue such as lack of expertise may induce an ambiguous attitude
(Petty & Cacioppo, 1986).
Lafferty and colleagues (2002) proposed a dual-credibility model of endorser and
corporate credibility to test the potential influence on people’s attitude toward messages from
different sources. Findings showed that endorser credibility has more direct effect on subjects’
attitude towards test ad than corporate. While corporate credibility has more comprehensive
effect on subjects’ attitude, such as attitude towards ad, brand and purchase intention, rather than
endorser celebrities mainly influenced attitude toward ads.
Based on this, the current study hypothesized:
H1: Participants would rate messages from human source (CEO) more favorably than
those from organization source (company).
H2: Participants would express greater favorability toward the company in crisis upon
exposure to messages from organization source (company) compared to those from
human source (CEO).
H3: Participants would express greater viral behavioral intentions toward messages from

19

organization source (company) compared to those from human source (CEO).
The current study also manipulated one aspect of virality related to the number of likes a
message has received. The number of likes, resembling a viral behavioral response, is expected
to enhance persuasion by increasing the sense of consensus to a message’s content. The
following section conceptualizes virality and provides the rationale for the study’s prediction in
relation to the persuasive effects of virality within the context of crisis communication.
Virality as a Message Feature
Alhabash and colleagues (2013a) argued that past literature yielded three different
approaches to defining virality. The first approach mainly focuses on access, and measures
virality by looking at how many people viewed a piece of content online in a given time period
(e.g., Bonchi, Castillo, and Ienco 2011; Guerini, Pepe, and Lepri 2011; Tucker 2011). The second
approach looks at the process by which a piece of content becomes viral through the lens of
electronic word-of-mouth (eWOM). eWOM is an extension of WOM, which refers to “oral,
person-to-person communication between a receiver and a communicator whom the receiver
perceives as non-commercial, regarding a brand, product, or service” (Arndt, 1967, p. 5). eWOM
offers the possibility for audience members to exchange information swiftly, multiple times and
anonymously, and further diminishes space and time limitations (Jasen, Zhang, Sobel, &
Chowhdury, 2009). Within this context, virality refers to the act of sharing content online. The
third approach defines virality as engagement and argues that users’ behaviors, including
accessing, liking/disliking, sharing, and commenting, collectively qualify engagement as a
measure of online advertising effectiveness (Tucker 2011, as cited in Alhabash et al., 2013a).
Alhabash and colleagues (2013a, 2013b) take a behavioral approach to defining virality.
Their definition takes a user-centered behavioral approach to identifying the “determinants of

20

user interactivity with a persuasive message disseminated and shared online” by looking at the
message’s affective evaluation, viral reach, and message deliberation, in addition to access
(Alhabash, 2013b, p. 4). Affective evaluation refers to users’ explicit emotional responses
expressed as behavioral responses to online content (e.g., liking a message on Facebook). Viral
reach deals with sharing of content online within a person’s network of friends (e.g., posting a
YouTube video on Facebook). Finally, message deliberation refers to users’ comments on a piece
of online content that could also entail affective evaluations of the content (e.g., posting a
comment on a friend’s Facebook status update).
Virality has become an essential metric to evaluating effective of online business-toconsumer communication. Companies do not make content popular. People popularize content.
The relevance of virality stems from consumers’ greater trust in information they see on their
immediate social network (Duana, Gub, &Whinston, 2008). A report by BrightEdge.com shows
that virality responses of liking and sharing messages online have become ever important. They
showed that the top 1,000 websites have been liked and shared over 65 million times on
Facebook. Additionally, the report showed that the “like” button is one of Facebook’s most used
features. ExactTarget (2009) reported that nearly half of respondents (45%) said they “like” a
company account on Facebook at least per month, with an average of liking 14 companies.
The number of Facebook likes can be understood in terms of audiences’ expectation. It
reflects an evaluation by others that may influence readers’ evaluations. The high number of likes
shows a majority sharing opinion with readers, which is related with reader’s expectation.
Expectation is defined as a belief about the probabilities associated with one’s future state of
affairs (Olson, 1996). Wilson, Lisle, Kraft and Wetzel (1989) raised affective expectation model
(AEM), which explained how affective expectations influence people’s affective reaction.

21

Researchers have identified several factors that influence people’s affective expectation (Jonas &
McGillis, 1976, Hoschschild, 1979): people’s previous personal experience, cultural norms and
other’s evaluation. Other’s evaluation is also called category-based expectation. It means that if
majorities of people evaluate one thing in a certain way, then other people may expect to
experience it in the same way. In this case, to readers, high number of likes brings more
compelling positive responses to messages than low number of likes.
Alhabash and colleagues (2013b) argued, using signaling theory (Krebs and Dawkins
1984), that the virality of online content could be a signal of the quality of the message, thus
increasing its persuasiveness. They found that highly liked messages resulted in greater attitudes
and viral behavioral intentions related to anti-cyberbullying messages shared on a Facebook page
of a mock non-profit organization. The current study will manipulate the number of likes for an
apology crisis response shared on Facebook. Participants will see the same message, yet half of
them will see it with a larger number of likes and the other half with a low number of likes. In
general, the study predicts that greater number of likes would result in greater persuasiveness of
the apology response. The study predicts that crisis response messages on Facebook with a high
number of likes will results in greater persuasive outcomes compared to those with low number
of likes. Thus, the study hypothesizes:
H4: Participants will express more favorable attitudes toward highly liked messages
compared to lowly liked ones.
H5: Participants will express more favorable attitudes toward the company in crisis
upon exposure to highly liked messages compared to lowly liked messages.
H6: Participants will express greater viral behavioral intentions toward highly liked
messages compared to lowly liked messages.

22

The study herein will also explore the effect of the interaction between virality (number
of likes) and source type (CEO vs. company page) on persuasiveness of the crisis response
message posted on Facebook. Due to limited research on this topic, the study reverts to asking
the following research questions:
RQ1: How would the interaction between the number of likes and source type affect
participants’ attitudes toward the message?
RQ2: How would the interaction between the number of likes and source type affect
participants’ attitudes toward the company in crisis?
RQ3: How would the interaction between the number of likes and source type affect
participants’ viral behavioral intentions?
Additionally, the current study investigated the effects of source type and virality on
perceived credibility. The ELM model discussed above, argues that perceived source credibility
is the sole outcome of persuasion. It also influences persuasive outcomes of attitude, behavioral
intention or actual behavior, therefore, to consider the effect of perceived source credibility. In
the current study, we explored the ways in which source type and virality affect the three facets
of perceived source credibility (attractiveness, trustworthiness, and expertise). Additionally, we
explored how each of these credibility types affects participants’ attitudes and viral behavioral
intentions. Thus, we raised the following three research questions:
RQ4: How do the three types of credibility predict participants’ attitude towards status
update?
RQ5: How do the three types of credibility predict participants’ attitude towards company
in crisis?
RQ6: How do the three types of credibility predict participants’ viral behavior intention?

23

METHOD
Design and Participants
This study used a 2 (source type: CEO vs. company) x 2 (number of likes: low vs. high) x
3 (message repetition) mixed factorial design. All factors except repetition were manipulated as
between-subjects. Participants (N = 268) for this experiment were recruited from large
introductory courses at the College of Communication Arts and Sciences at Michigan State
University and were rewarded with extra/course credit in exchange of participation. Nearly six in
10 participants were female (59.1%). The majority of participants reported they were
white/Caucasian (70.1%), followed by Asian (16.0%), African American (10.1%), Hispanic or
Latino (1.1%), others (2.7%). The age of respondents ranged from 19 to 27, with an average of
21 (M=21 SD=1.64).
Independent Variables
Source Type. Source type was manipulated between subjects. It was operationalized as
the type of source appearing as the owner of the Facebook profile. Participants were randomly
assigned to view messages posted a CEO’s or a company’s Facebook page. Profile pictures for
CEOs were pretested on valence, arousal, and attractiveness with comparable pictures included
in the stimuli construction. Company logos were designed and pretested specifically for this
study.
Number of Likes. Message virality was manipulated between subjects. It is
operationalized as the number of “likes” printed beneath the crisis response message.
Participants were randomly assigned to one of two levels: low likes vs. high likes. A pilot study
has been done to determine the number of likes. Five recent Facebook posts from top 20 most

24

“liked” companies (Felix, 2012) in 2012 were chosen as sample. After calculating the mean,
maximum (19) and minimum number (118,296) of likes of each post, we decided to randomly
pick a number from 10-30 as low likes, and a number from 110,000 to 130,000 as high likes.
Repetition. Each of the four treatment levels was represented by three repetitions.
Dependent variable
Attitude toward Crisis Response Message. Attitude toward crisis response message
was measured using past measures of attitude toward message (Mackenzie, Lutz & Belch, 1986).
Participants rated each crisis response message using three 7-point semantic differential scales:
bad/good, negative/positive, unfavorable/favorable. Upon satisfactory factor and reliability
analyses (see Table 2), items were aggregated into single variables corresponding with each
message.
Table 2.
Factor analysis and reliability for attitude toward crisis response message.
Crisis Case 1
Crisis Case 2
Crisis Case 3
Item
Mean SD Loading Mean SD Loading Mean SD Loading
Positive/Negative
4.77 1.50 0.75
4.60 1.62 0.78
4.71 1.60 0.74
Bad/Good
4.68 1.53 0.87
4.57 1.63 0.93
4.79 1.51 0.88
Unfavorable/Favorable
4.74 1.52 0.91
4.55 1.59 0.90
4.70 1.55 0.90
Eigenvalues
2.14
2.27
2.13
% Of Variance
71.47
75.63
71.13
Reliability
0.88
0.90
0.88
Attitude toward Company in Crisis. Attitude toward the company were measured
using past measures of attitude towards the brand (Mackenzie, et al, 1986). Participants rated
each Facebook message using three 7-point semantic differential scales: bad/good,
negative/positive, unfavorable/favorable. Upon satisfactory factor and reliability analyses (see
Table 3), items were aggregated into single variables corresponding with each message.

25

Table 3.
Factor analysis and reliability for attitude toward company in crisis.
Crisis Case 1
Crisis Case 2
Item
Mean SD Loading
Mean SD Factor
Good/Bad
4.53 1.590 0.835
4.24 1.662 0.922
Negative/Positive
4.55 1.558 0.969
4.24 1.661 0.971
Unfavorable/Favorable
4.54 1.585 0.939
4.18 1.695 0.914
Eigenvalues
2.52
2.63
% Of Variance
83.89
87.65
Reliability
0.94
0.96

Crisis Case 3
Mean SD Loading
4.50 1.640 0.850
4.47 1.625 0.937
4.42 1.631 0.976
2.55
85.14
0.94

Viral Behavioral Intentions (VBI). To measure viral behavioral intentions, the study
used items validated by Alhabash et al. (2013b). Participants rated their agreement/disagreement
to five statements using 7-point Likert type scales anchored by “Strongly Agree” and “Strongly
Disagree”. Upon satisfactory factor and reliability analyses (see Table 3), items were aggregated
into single variables corresponding with each message.
Table 4.
Factor analysis and reliability for attitude toward viral behavior intention.
Crisis Case 1
Crisis Case 2
Crisis Case 3
Item
Mean SD Loading Mean SD Loading Mean SD Loading
The status update is worth
2.83 1.73 0.75
2.82 1.70 0.78
2.89 1.81 0.80
sharing with others
I would recommend this
2.42 1.64 0.79
2.44 1.50 0.87
2.55 1.69 0.82
status update to others
I would “like” this status
2.74 1.84 0.68
2.68 1.77 0.65
2.62 1.75 0.74
update
I would “share” this status
2.17 1.51 0.92
2.22 1.52 0.89
2.23 1.53 0.89
update
I would “comment” on
1.93 1.33 0.65
2.01 1.29 0.67
2.07 1.51 0.72
this status update
Eigenvalues
2.92
3.01
3.16
% Of Variance
58.40
60.18
63.28
Reliability
0.87
0.87
0.89
Perceived source credibility-attractiveness. To measure attractiveness, this study used
items validated by Ohanian (1990). Participants rated their agreement/disagreement to five
statements using 7- point Likert type scales anchored by “Strongly Agree” and “Strongly
26

Disagree”. Upon satisfactory factor and reliability analyses (see Table 3), items were aggregated
into single variables corresponding with each message.
Table 5.
Factor analysis and reliability for attractiveness.
Crisis Case 1
Item
Mean SD Loading
Attractive/Unattractive
3.72 1.21 0.82
Classy/Not classy
4.31 1.36 0.57
Beautiful/Ugly
3.73 1.01 0.73
Elegant/Plain
3.55 1.41 0.53
Sexy/Not sexy
2.77 1.38 0.49
Eigenvalues
2.05
% Of Variance
40.91
Reliability
0.75

Crisis Case 2
Mean SD Loading
3.78 1.23 0.79
4.16 1.39 0.58
3.69 1.06 0.79
3.56 1.42 0.55
2.95 1.43 0.46
2.10
41.93
0.75

Crisis Case 3
Mean SD Loading
3.88 1.20 0.74
4.40 1.40 0.58
3.73 1.04 0.79
3.68 1.37 0.63
2.78 1.43 0.34
2.02
40.40
0.93

Perceived source credibility-trustworthiness. To measure trustworthiness, this study
used items validated by Ohanian (1990). Participants rated their agreement/disagreement to five
statements using 7- point Likert type scales anchored by “strongly agree” and “strongly
disagree”. Upon satisfactory factor and reliability analyses (see Table 3), items were aggregated
into single variables corresponding with each message.
Table 6.
Factor analysis and reliability for trustworthiness.
Crisis Case 1
Item
Mean SD Loading
Dependable/Not
4.54 1.415 0.814
dependable
Honest/Dishonest
4.93 1.467 0.859
Reliable/Unreliable
4.55 1.476 0.867
Sincere/Insincere
4.85 1.574 0.852
Trustworthy/Untrustworthy 4.66 1.397 0.865
Eigenvalues
3.63
% Of Variance
72.51
Reliability
0.929

27

Crisis Case 2
Mean SD Loading

Crisis Case 3
Mean SD Loading

4.46 1.484 0.786

4.55 1.386 0.826

4.88
4.53
4.67
4.54

5.02
4.69
4.87
4.73

1.458
1.505
1.496
1.433
3.34
66.82
0.909

0.782
0.836
0.811
0.869

1.375
1.432
1.460
1.440
3.44
68.72
0.916

0.738
0.845
0.860
0.868

Perceived source credibility-expertise. To measure expertise, this study used items
validated by Ohanian (1990). Participants rated their agreement/disagreement to 5 statements
using 7- point Likert type scales anchored by “strongly agree” and “strongly disagree”. Upon
satisfactory factor and reliability analyses (see Table 3), items were aggregated into single
variables corresponding with each message.
Table 7.
Factor analysis and reliability for expertise.
Crisis Case 1
Item
Mean SD Loading
Expert/Not expert
4.53 1.38 0.83
Experienced/Inexperienc
4.70 1.40 0.87
ed
Knowledgeable/Unknow
4.88 1.33 0.86
ledgeable
Qualified/Not qualified
4.77 1.38 0.88
Skilled/Unskilled
4.56 1.38 0.85
Eigenvalues
3.77
% Of Variance
75.39
Reliability
0.94

Crisis Case 2
Mean SD Loading
4.41 1.37 0.83

Crisis Case 3
Mean SD Loading
4.56 1.33 0.83

4.54 1.40 0.84

4.69 1.36 0.87

4.52 1.42 0.89

4.77 1.29 0.85

4.55 1.44 0.89
4.39 1.34 0.86
3.73
74.65
0.94

4.81 1.34 0.89
4.60 1.33 0.89
3.75
75.09
0.94

Stimuli and Pretest
In the main test, participants were exposed to a total of three crisis scenarios, followed by
a screen shot for a crisis response. Crisis scenarios and the content of the apology crisis response
were identical across all conditions. Participants were randomly assigned to one of the following
four conditions: CEO post with high likes, CEO post with low likes, company post with high
likes, and company post with low likes. All materials were constructed specifically for this
experiment and were pretested with a smaller sample of participants, other than those of the main
experiment.
To construct the stimuli, the study pretested the following elements: the crisis scenario
and the profile picture for CEO or company logo. First, the three crisis scenarios used in the
28

main experiment were selected from a pool of 12 different scenarios for crises faced by fictitious
companies of similar trade orientation. Participants (N = 20) rated each of the 12 scenarios using
7-point scales for the scenario’s level of arousal, positivity, negativity, and crisis severity. We
selected three cases with comparable arousal, positivity, negativity, and crisis severity ratings
(see Table 5). The ANOVA results showed that there was no significant difference among these
cases (Arousal: F(1, 20)=0.02, ns; Positivity: F(1, 20)=0.56, ns; Negativity: F(1, 20)=0.00, ns;
Severity: F(1, 20)=0.35, ns).
Table 8.
Descriptive statistics of pre-selected crisis cases.
Arousal
Positivity
Case No.
Mean SD
Mean SD
6
4.65 1.53
2.50 1.00
8
4.90 1.12
2.45 1.23
11
4.70 1.56
2.30 1.03

Negativity
Mean SD
5.50 0.83
5.70 0.98
5.50 1.10

Severity
Mean SD
5.60 0.88
5.40 1.23
5.45 1.23

Second, to select the profile pictures that accompany the Facebook posts, we pretested
the images of CEO and images of company logo. The three CEO pictures were selected from a
pool of 12 pictures of male CEOs that was pretested using 7-point scales for attractiveness,
trustworthiness and expertise. According to a CEO survey conducted by Inc.com website, 68%
of CEOs of fortune 500 companies fell into age 35-55, 90% are male, and 79% are Caucasian
(Inc. 2012). Therefore, in this study, CEO images were limited to middle-aged, Caucasian male.
The three logo pictures were also selected from a pool of 12 pictures of logo that was pretested
using 7-point scales for attractiveness, trustworthiness and expertise. We chose 3 logos and 3
CEO images that attractiveness, trustworthiness and expertise which were in median level in this
test, and the ANOVA results showed there was no significant difference among these logos
(Attractiveness: F(1, 20)=0.05, ns; Trustworthiness: F(1, 20)=0.02, ns; Expertise: F(1, 20)=0.23,

29

ns), and CEO images (Attractiveness: F(1, 20)=4.17, ns; Trustworthiness: F(1, 20)=0.03, ns;
Expertise: F(1, 20)=0.35, ns). Hence, the manipulation on CEO images or company logos was
verified.
Table 9.
Descriptive statistics of pre-test selected CEO images and company logos.
Attractiveness
Trustworthiness
Item
Mean SD
Mean SD
CEO image 1
3.19 1.08
4.76 1.41
CEO image 2
3.86 1.28
4.05 1.47
CEO image 3
3.86 1.42
4.81 0.81
F value
4.17
0.03
Company logo 1
4.00 1.79
4.00 1.58
Company logo 2
3.95 1.12
3.81 1.25
Company logo 3
3.90 1.58
4.05 1.32
F value
0.05
0.02

Expertise
Mean SD
5.10 1.09
4.90 0.94
4.95 0.67
0.35
4.10 1.55
3.95 1.50
3.90 1.45
0.23

Upon completion of the pretest, comparable scenarios, CEO and Logo images were
selected for the construction of the three sets of stimuli. The crisis scenarios were presented to
participants as text descriptions. Each scenario was matched with the same standard crisis
response. The same crisis response was also matched with a picture of a CEO or a company
logo. Additionally, both CEO and company posts crossed with low and high number of likes
with 3 repetitive cases (see Figure 1 to Figure 12).
Main Study Procedure
The researcher recruited participants from two channels: parts of student participants took
this survey via College Arts and Sciences experiment management system (SONA). Parts of
student participants were recruited from introductory classes at the College of Communication
Arts and Sciences. These two parts of data does not have significant difference between each
1

other. Students received a recruit email with this survey link in it, and they were randomly
assigned to the four treatment conditions: (1) CEO post with high likes, (2) CEO post with low
30

likes, (3) company post with high likes, and (4) company post with low likes. Participants
completed the study online using the Qualtrics online survey software. Upon receiving the
experiment link, participants were asked to carefully read and electronically sign an informed
consent form. Upon providing consent, the online system randomly assigned participants to one
of the four treatment conditions. Participants, then, were given instructions about the experiment.
Participants answered a few questions about their Facebook uses and motivations. Afterwards,
they were exposed to the three crisis scenarios (order counterbalanced). Each crisis scenario was
followed by the matching crisis response, followed by questions about attitudes and behavioral
intentions. Participants reported their demographic information (age, gender, ethnicity,
educational level, and income level). Finally, participants were asked to indicate their names and
course name or 4 digits generated from SONA system if they wish to acquire extra/course credit.
Manipulation Check
Because the images of CEO or logos of companies are easy to identify for participants,
manipulation check was only conducted for the number of likes. Participants, after exposure to
each message, were asked: “How do you consider this Facebook status update as low likes/high
likes” (see Table 8). The Chi-square analysis for each message shows that the majority of
participants identified status updates with low number of likes and high number of likes
accurately per manipulation. Hence, the manipulation was verified in this study.

31

Table 10.
Manipulation check on number of likes
Case
Number of likes
% Low likes

% High likes

High like condition
Low like condition
High like condition
Low like condition
High like condition
Low like condition

74.1%
25.9%
80.3%
19.7%
80.2%
19.8%

Case 1
Case 2
Case 3

28.2%
71.8%
27.2%
72.8%
26.0%
74.0%

2

df

X

Sig. (p)

1

54.44

<0.01

1

72.73

<0.01

1

75.67

<0.01

Data Reduction and Analysis
Items for attitudes toward crisis response, attitudes toward the company, and viral
behavioral intentions were submitted to confirmatory factor analyses and reliability analyses for
each crisis scenario. Upon satisfactory factor and reliability analyses, items per construct were
aggregated per scenario. To test H1, H2, and H3, data for attitudes toward the crisis response,
attitudes toward the company, and viral behavioral intentions, respectively, were submitted to 2
(source type) x 3 (repetition) repeated measures analysis of variance (ANOVA). To test H4, H5,
and H6, data for attitudes toward the crisis response, attitudes toward the company, and viral
behavioral intentions, respectively, were submitted to 2 (number of likes) x 3 (repetition)
repeated measures analysis of variance (ANOVA). To answer RQ1, RQ2, and RQ3, data for
attitudes toward the crisis response, attitudes toward the company, and viral behavioral
intentions, respectively, were submitted to 2 (source type) x 2 (number of likes) x 3 (repetition)
repeated measures analysis of variance (ANOVA). Finally, to answer R4, R5 and R6 we ran
linear regression models with the following predictors: attractiveness, trustworthiness and
expertise, with attitude towards status update, company in crisis and viral behavior intention as
criterion variables.

32

RESULTS
This study investigated the effects of message virality and source type on the
persuasiveness of crisis response messages of fictitious companies on Facebook. Some
descriptive results are reported before reporting the results of our hypothesis and research
questions,
Descriptive Results
The majority of participants reported that they have a Facebook account (97.8%). On
average, participants reported having 754 Facebook friends (M = 754.82, SD = 517.87). They
also reported that they, on average, daily use Facebook for over two hours (M = 122.53 minutes,
SD = 289.96). Respondents also reported average 27 commercial companies or brands they
interact with (M = 26.76, SD=44.50),
In the following section, we test the hypothesis and answer research questions by
reporting results related to the effects of virality and source type on persuasiveness of online
crisis responses on Facebook. To ensure ease of results presentation, results are grouped as a
function of dependent variable. The first set corresponds with H1, H4, and RQ1, deals with the
main effects of source type and the number of likes, and the interaction between them on
attitudes toward the status update. Next, we report main and interaction effects on attitudes
toward the company (H2, H5, and RQ2). Finally, we report main and interaction effects on viral
behavioral intentions (H3, H6, and RQ3).
Attitude toward Status Update
H1 predicted that participants would rate messages from human source more favorably
than those from organization source. H4 predicted that participants would express more
favorable attitudes toward highly liked messages compared to lowly liked ones. The first
research question addressed the interaction effects of source type and virality on attitude towards
33

the status update. To test H1 and H4, and answer RQ1, data for attitudes toward the status update
were submitted to a 2 (affective evaluation: low likes vs. high likes) x 2 (source type: CEO vs.
Company) x 3 (message repetition) repeated measures analysis of variance (ANOVA), with
repeated measures on the last factor. Results illustrated the main effects of source type (F(1, 233)
= 0.05, ns) and the number of likes (F(1, 233) = 0.048, ns) on attitudes toward status updates
were not significant. Additionally, the effect of the interaction between source type and virality
on subjects’ attitude toward status update was not significant (F(1, 233) = 0.19, ns) (see Tables
11 and 12).
Table 11.
Tests of between-subject effects on attitude towards status update
Variables
Source
Like
Source X Likes

F

df

0.53
0.48
0.19

1
1
1

2
η p
0.00
0.00
0.00

Sig.
0.82
0.83
0.66

Table 12.
Mean for attitudes towards status update
Variables
Low likes
Mean
SD
CEO
4.74
0.14
Company
4.64
0.14

High likes
Mean
4.64
4.67

SD
0.14
0.20

Attitude toward Company in Crisis
H2 predicted that participants would express more favorable attitudes toward the
company in crisis when the status update is posted by human source than organization source.
H5 predicted that participants would express more favorable attitudes toward the company in
crisis upon exposure to highly liked messages compared to lowly liked messages. The second
research question addressed that the interaction effects of source type and virality on attitude

34

towards the crisis company. To test H2, H5, and RQ2, data were submitted to a 2 (affective
evaluation: low likes vs. high likes) x 2 (source type: CEO vs. Company) x 3 (message repetition)
repeated measures analysis of variance (ANOVA), with repeated measures on the last factor.
Results, illustrated that the main effect of source type (F(1, 233) = 0.084, ns) and number of likes
(F(1, 233) = 0.23, ns), and the interaction between source type and the number of likes (F(1, 233)
= 1.08, ns) on attitudes toward the company were not significant (see Tables 13 and 14).
Table 13.
Tests of between-subject effects on attitude towards company in crisis
Variables
Source
Like
Source X Likes

F

df

0.08
0.23
1.08

1
1
1

2
η p
0.00
0.00
0.00

Sig.
0.77
0.63
0.30

Table 14.
Mean for attitudes towards company in crisis
Variables
Low likes
Mean
SD
CEO
4.46
0.15
Company
4.64
0.14

High likes
Mean
4.36
4.67

SD
0.15
0.20

Viral Behavior Intentions
H3 predicted that participants would express greater viral behavioral intentions toward
messages from human source compared to those from organization source. H6 predicted that
participants would express greater viral behavioral intentions toward highly liked messages
compared to lowly liked messages. The third research question addressed that the interaction
effects of source type and virality on attitude towards viral behavior intention. To test H3, H6,
and RQ3, data were submitted to a 2 (affective evaluation: low likes vs. high likes) x 2 (source
type: CEO vs. Company) x 3 (message repetition) repeated measures analysis of variance

35

(ANOVA), with repeated measures on the last factor. Results illustrated that the main effects of
source type (F(1, 232) = 0.519, ns) and number of likes (F(1, 232) = 0.62, ns), as well as the
effect of interaction between source type and the number of likes (F(1, 232) < .01, ns) on viral
behavioral intentions was not significant (see Tables 15 and 16).
Table 15.
Tests of between-subject effects on viral behavior intention
Variables
Source
Like
Source X Like

F

df

0.52
0.52
0.04

1
1
1

2
η p
0.00
0.00
0.00

Sig.
0.47
0.43
0.95

Table 16.
Mean results for viral behavior intention
Variables
Low likes
Mean
SD
CEO
2.48
0.15
Company
2.61
0.15

High likes
Mean
2.36
2.47

SD
0.15
0.21

Predicting Attitudes and VBI from Perceived Source Credibility
The last set of research questions (RQ4-RQ6) asked about the ways in which the three
types of perceived source credibility (attractiveness, trustworthiness, and expertise) predicted
attitudes toward status update, attitudes toward the company, and viral behavioral intentions, as a
function of source type, virality, and the interaction between them. To answer these research
questions, a set of linear regression models were run: (1) with all the data and without
considering effects of the treatment conditions (source type and number of likes); (2) by splitting
the data set by source type; (3) by splitting the data set by the number of likes; and (4) by
splitting the data set by source type and number of likes (interaction).

36

Perceived source credibility and Attitude toward Status Update
RQ4 addressed how the three types of credibility would predict participants’ attitudes
towards status update. To answer RQ4, a set of linear regression models was run with the
following predictors: attractiveness, trustworthiness, and expertise, and with attitudes toward
status update as a criterion variable. First, the model was run with all of the data regardless of
conditions. Results indicated that the regression models was statistically significant (R = 0.67,
2

adjusted R =0.44, F(3, 265) = 70.44, p < .01). Results showed that trustworthiness was the
strongest predictor of participants’ attitudes towards status update (β = 0.49, t(265) = 5.32, p
< .01). Source expertise was marginally significant in predicting attitudes toward status updates
(β = 0.18, t(265) = 1.89, p = .06). Attractiveness was not a significant predictor of attitudes
toward status updates (β = 0.26, t(265) = 0.46, ns).
Table 17.
Summary of linear regression analysis for attitude towards status update (overall data)
Predictors
B
SE
β
t
Sig. (p)
Attractiveness
0.04
0.09
0.26
0.46
0.65
Trustworthiness
0.54
0.10
0.49
5.32
0.00
Expertise
0.21
0.11
0.18
1.89
0.06
2
Model Statistics
R = 0.67, adjusted R = 0.44, F(3, 265) = 70.44, p< .01

Second, the same regression model was run by splitting data by source type. Results
indicated that the regression model was significant for both the CEO condition (R = 0.72,
2

adjusted R = 0.51, F(3, 131) = 46.49, p < .01) and the organization condition (R = 0.62, adjusted
2

R = 0.37, F(3, 130) = 26.97, p < .01). Trustworthiness was the strongest predictor of
participants’ attitude towards status update in both CEO condition (β = 0.45, t(131) = 3.00, p <
.01) and organization group (β = 0.49, t(131) = 4.16, p < .01). In both conditions, as shown in
37

Table 18, neither attractiveness nor expertise were significant predictors of attitudes toward
status updates.
Table 18.
Summary of linear regression analysis for attitude towards status update (spilt file by “source”)
B

SE

β

t

Sig. (p)

0.18
0.46
0.22

0.12
0.15
0.16

0.11
0.45
0.21

1.48
3.00
1.35

0.14
0.00
0.17

Model Statistics
R = 0.72, adjusted R = 0.51, F(3, 131) = 46.49, p< .01
Attractiveness
-0.07
0.13
-0.05
-0.57
Organization Trustworthiness
0.59
0.14
0.49
4.16
Expertise
0.23
0.16
0.18
1.46

0.57
0.00
0.15

Source
CEO

Predictors
Attractiveness
Trustworthiness
Expertise
2

Model Statistics

2

R = 0.62, adjusted R = 0.37, F(3, 130) = 26.97, p< .01

Third, the same model was run by splitting data by number of likes. Results indicated that
2

the regression model was significant for both high likes condition (R = 0.61, adjusted R = 0.36,
2

F(3, 131) = 25.84, p < .01) and low likes condition (R = 0.74, adjusted R = 0.54, F(3, 130) =
52.60, p < .01). Based on the result, expertise was a significant predictor in high likes group (β =
0.42, t(131) = 2.91, p < .01), and trustworthiness was a significant predictor low likes group (β =
0.76, t(131) = 6.47, p < .01). None of the other credibility types were significant predictors of
attitudes toward status updates.

38

Table 19.
Summary of linear regression analysis for attitude towards status update (spilt file by “likes”)
Likes
Predictors
B
SE
β
t
Sig. (p)
Attractiveness
0.11
0.13
0.07
0.86
0.39
High
Trustworthiness
0.20
0.16
0.17
1.22
0.22
Expertise
0.50
0.17
0.42
2.91
0.00
2
Model Statistics
R = 0.61, adjusted R = 0.36, F(3, 131) = 25.84, p< .01
Attractiveness
-0.04
0.11
-0.03
-0.38
0.70
Low
Trustworthiness
0.83
0.13
0.76
6.47
0.00
Expertise
-0.01
0.14
-0.01
-0.05
0.96
2
Model Statistics
R = 0.74, adjusted R = 0.54, F(3, 130) = 52.60, p< .01

Lastly, the same regression model was run by splitting data by source type and number of
likes (interaction). Results showed that the regression model was significant for CEO and high
2

likes (R = 0.61, adjusted R = 0.35, F(3, 64) = 12.89, p < .01), CEO and low likes group (R =
2

0.81, adjusted R = 0.64, F(3, 63) = 39.48, p < .01), organization and high likes (R = 0.62,
2

adjusted R = 0.36, F(3, 63) = 13.17, p < .01), and organization and low likes (R = 0.69, adjusted
2

R = 0.46, F(3, 63) = 19.33, p < .01), With regards to status updates posted by a CEO and had a
low number of likes, trustworthiness significantly predicted attitudes toward status updates (β =
0.62, t(64) = 3.16, p < .01), while none of the credibility types predicted attitudes toward status
updates when the CEO post had a high number of likes. With regards to the status updates posted
by an organization, expertise was a significant predictor when the update had high likes (β =
0.57, t(63) = 2.86, p = .01), while trustworthiness was a significant predictor when the status
update had low likes ( β = 0.79, t(63) = 5.35, p < .01).

39

Table 20.
Summary of linear regression analysis for attitudes toward status update (spilt file by “likes”
and “source”)
Source
Likes
Predictors
B
SE
β
t
Sig. (p)
Attractiveness
0.17
0.21
0.10
0.83
0.41
CEO
High Trustworthiness
0.34
0.24
0.33
1.42
0.16
Expertise
0.27
0.25
0.24
1.07
0.29
2
Model Statistics
R = 0.61, adjusted R = 0.35, F(3, 64) = 12.89, p< .01
Attractiveness
0.16
0.15
0.11
1.05
0.30
CEO
Low Trustworthiness
0.63
0.20
0.62
3.16
0.00
Expertise
0.12
0.21
0.12
0.57
0.57
Model Statistics

2

R = 0.81, adjusted R = 0.64, F(3, 63) = 39.48, p< .01
Attractiveness
0.01
0.18
0.01
Organization High Trustworthiness
0.06
0.24
0.05
Expertise
0.77
0.27
0.57
2
Model Statistics
R = 0.62, adjusted R = 0.36, F(3, 63) = 13.17, p< .01
Attractiveness
-0.24
0.17
-0.16
Organization Low Trustworthiness
0.93
0.17
0.79
Expertise
-0.03
0.19
-0.03
Model Statistics

0.07
0.27
2.86

0.95
0.79
0.01

-1.39
5.35
-0.17

0.17
0.00
0.87

2

R = 0.69, adjusted R = 0.46, F(3, 63) = 19.33, p< .01

Perceived source credibility and attitude toward Company in Crisis
RQ5 addressed that how the three types of credibility would predict participants’ attitude
towards company in crisis. To answer RQ5, a set of linear regression models were run with
following predictors: attractiveness, trustworthiness, and expertise, and with attitude toward
status update as a criterion variable. First, model was run with all of the data regardless
2

conditions. The regression model was significant (R = 0.68, adjusted R = 0.46, F(3, 265) =
76.13, p < .01). Trustworthiness was the strongest and only significant predictor of attitudes
towards status update (β = 0.52, t(265) = 5.75, p < .01).

40

Table 21.
Summary of linear regression analysis for attitude towards company in crisis (overall data)
Predictors
B
SE
β
t
Sig. (p)
Attractiveness
0.16
0.10
0.09
1.68
0.09
Trustworthiness
0.63
0.11
0.52
5.75
0.00
Expertise
0.15
0.12
0.12
1.28
0.20
2
Model Statistics
R = 0.68, adjusted R = 0.46, F(3, 265) = 76.13, p< .01

Second, model was run by splitting data by source type. The regression mode was
2

significant for CEO condition (R = 0.73, adjusted R = 0.53, F(3, 131) = 42.84, p < .01) and
2

organization condition (R = 0.64, adjusted R = 0.39, F(3, 130) = 29.14, p < .01). When the
status update was posted by a CEO, attitudes toward the company was significantly predicted by
trustworthiness (β = 0.73, t(131) = 4.97, p < .01) and attractiveness (β = 0.18, t(131) = 2.35, p <
.05). When the update was posted by the organization, trustworthiness (β = 0.39, t(131) = 3.34, p
< .01) and expertise ( β = 0.28, t(131) = 2.24, p < .05) significantly predicted attitudes toward the
company.
Table 22.
Summary of linear regression analysis for attitude towards company in crisis (spilt file by
“source”)
Source
Predictors
B
SE
β
t
Sig. (p)
Attractiveness
0.32
0.14
0.18
2.35
0.02
CEO
Trustworthiness
0.85
0.17
0.73
4.97
0.00
Expertise
-0.14
0.18
-0.12
-0.80
0.42
2
Model Statistics
R = 0.73, adjusted R = 0.53, F(3, 131)= 42.84, p< .01
Attractiveness
0.02
0.13
0.01
0.12
0.91
Organization Trustworthiness
0.50
0.15
0.39
3.34
0.00
Expertise
0.37
0.17
0.28
2.24
0.02
2
Model Statistics
R=0.64, adjusted R = 0.39, F(3, 130) = 29.14, p< .01

41

Third, the same regression model was run by splitting data by number of likes. The
2

regression model was significant for high likes (R = 0.61, adjusted R = 0.36, F(3, 131) = 25.84,
2

p < .01) and low likes (R = 0.74, adjusted R = 0.54, F(3, 130) = 52.60, p< .01). When the status
update had high likes, expertise (β = 0.33, t(131) = 2.25, p < .05) was a significant predictor of
attitudes toward the company, and marginally by trustworthiness (β = 0.25, t(131) = 1.77, p =
.08). Conversely, then status updates had low likes, trustworthiness was the strongest and only
predictor of attitudes toward the company (β = 0.76, t(131) = 6.77, p < .01).
Table 23.
Summary of linear regression analysis for attitude towards company in crisis (spilt file by
“likes”)
Likes
Predictors
B
SE
β
t
Sig. (p)
Attractiveness
0.14
0.15
0.07
0.92
0.36
High
Trustworthiness
0.32
0.18
0.25
1.77
0.08
Expertise
0.44
0.20
0.33
2.25
0.03
2
Model Statistics
R = 0.60, adjusted R = 0.34, F(3, 131) = 24.27, p< .01
Attractiveness
0.17
0.12
0.11
1.43
0.16
Low
Trustworthiness
0.88
0.13
0.76
6.77
0.00
Expertise
-0.7
0.14
-0.06
-0.50
0.62
2
Model Statistics
R = 0.77, adjusted R = 0.59, F(3, 130) = 63.70, p< .01

Lastly, the same regression model was run by splitting data by source type and number of
likes (interaction). Results showed that the regression model was significant for CEO and high
2

likes (R = 0.62, adjusted R = 0.35, F(3, 64) = 13.20, p < .01), CEO and low likes group (R =
2

0.86, adjusted R = 0.72, F(3, 63) = 58.82, p < .01), organization and high likes (R = 0.60,
2

adjusted R = 0.32, F(3, 63) = 11.50, p < .01), and organization and low likes (R = 0.70, adjusted
2

R = 0.47, F(3, 63) = 20.47, p < .01), With regards to status updates posted by a CEO and had a

42

low number of likes, trustworthiness significantly predicted attitudes toward company in crisis (β
= 0.79, t(63) = 5.35, p < .01), followed by attractiveness (β = 0.22, t(63) = 2.49, p < 0.05) and
expertise (β = -0.46, t(63) = -2.55, p = .01). With regards to the status updates posted by an
organization, expertise was a significant predictor when the update had high likes (β = 0.53,
t(64) = 2.58, p = .01), while trustworthiness was a significant predictor when the status update
had low likes (β =0.59, t(63) =4.08, p<.01).
Table 24.
Summary of linear regression analysis for attitudes toward company in crisis (spilt file by
“likes” and “source”)
Source
Like
Predictors
B
SE
β
t
Sig. (p)
Attractiveness
0.30
0.25
0.14
1.20
0.24
CEO
High Trustworthiness
0.54
0.29
0.44
1.88
0.07
Expertise
0.15
0.30
0.11
0.49
0.63
Model Statistics
CEO

Low

Model Statistics
Organization

Low

Model Statistics

2.49
6.51
-2.55

0.02
0.00
0.01

-0.13
0.44
2.58

0.90
0.66
0.01

-0.05
4.08
0.89

0.96
0.00
0.38

2

R = 0.86, adjusted R = 0.72, F(3, 63) = 58.82, p< .01
Attractiveness
-0.03
0.19
-0.02
High Trustworthiness
0.11
0.26
0.08
Expertise
0.75
0.29
0.53

Model Statistics
Organization

2

R = 0.62, adjusted R = 0.35, F(3, 64) = 13.20, p< .01
Attractiveness
0.35
0.14
0.22
Trustworthiness
1.24
0.19
1.12
Expertise
-0.51
0.20
-0.46

2

R = 0.60, adjusted R = 0.32, F(3, 63) = 11.50, p< .01
Attractiveness
-0.01
0.18
-0.01
Trustworthiness
0.74
0.18
0.59
Expertise
0.18
0.20
0.14
2

R = 0.70, adjusted R = 0.47, F(3, 63) = 20.47, p< .01

Perceived source credibility and Viral Behavior Intention
RQ6 addressed that how the three types of credibility would predict viral behavior
intention. To answer RQ6, a set of linear regression models was conducted with the following
predictors: attractiveness, trustworthiness, and expertise, and with attitude toward status update

43

as a criterion variable. First, model was run with all of the data regardless conditions. The
2

regression model was not significant (R=0.15, Adjusted R =0.01, F(3, 265)=1.98, ns).
Table 25.
Summary of linear regression analysis for viral behavior intention (overall data)
Predictors
B
SE
β
t
Sig. (p)
Attractiveness
0.23
0.12
0.14
1.90
0.06
Trustworthiness
0.01
0.14
0.01
0.06
0.95
Expertise
0.01
0.15
0.00
0.02
0.98
2
Model Statistics
R = 0.15, adjusted R = 0.01, F(3, 265) = 1.98, p = .12

Second, the same model was run by splitting data by source type. The regression was
2

neither significant for the CEO group (R=0.16, adjusted R =0.00, F(3, 131)= 1.07, ns) or the
2

organization group (R=0.18, adjusted R =0.01, F(3, 130)=1.56, ns).
Table 26.
Summary of linear regression analysis for viral behavior intention (spilt file by “source”)
Source
Predictors
B
SE
β
t
Sig. (p)
Attractiveness
0.26
0.18
0.16
1.44
0.15
CEO
Trustworthiness
0.19
0.23
0.18
0.84
0.40
Expertise
-0.24
0.24
-0.22
-1.02
0.31
2

Model Statistics
R = 0.16, adjusted R
Attractiveness
Organization Trustworthiness
Expertise
Model Statistics

= 0.00, F(3, 131) = 1.07, p = .36
0.18
0.16
0.11
1.09
-0.09
0.19
-0.07
-0.46
0.10
0.20
0.15
0.92

0.28
0.65
0.36

2

R = 0.18, adjusted R = 0.01, F(3, 130) = 1.56, p = .20

Third, model was run by splitting data by number of likes. The result indicated there were
2

no significant predictor in either high likes group (R=0.14, adjusted R =-0.00, F(3, 131)=0.92, ns)
2

or low likes group (R=0.16, adjusted R =0.00, F(3, 130)=1.11, ns).

44

Table 27.
Summary of linear regression analysis for viral behavior intention (spilt file by “likes”)
Likes

Predictors

B

SE

β

t

Attractiveness
0.27
0.17
0.15
1.54
Trustworthiness
0.07
0.21
0.06
0.33
Expertise
0.44
0.23
-0.08
-0.47
2
Model Statistics
R = 0.14, adjusted R = 0.00, F(3, 131) = 0.92, p = .44
Attractiveness
0.17
0.17
0.12
1.02
Low
Trustworthiness
-0.03
0.19
-0.03
-0.14
Expertise
0.09
0.20
0.08
0.43
High

Sig. (p)
0.13
0.74
0.64
0.31
0.89
0.67

2

Model Statistics

R = 0.16, adjusted R = 0.00, F(3, 130) = 1.11, p = .35

Lastly, the model was run by splitting data by source type and number of likes
(interaction). The result indicated there were no significant predictor in either CEO and high
2

likes group (R = 0.20, adjusted R = 0.00, F(3, 64) = 0.92, ns), CEO and low likes group
2

(R=0.18, adjusted R =-0.02, F(3, 63)=0.67, ns), organization and high likes group (R = 0.22,
2

2

adjusted R =0.00, F(3, 63)=1.02, ns) or organization and low likes group (R = 0.25, adjusted R
= 0.02, F(3, 63) =1.43, ns).

45

Table 28.
Summary of linear regression analysis for viral behavior intention (spilt file by “likes” and
“source”)
Source
Like
Predictors
B
SE
β
t
Sig. (p)
Attractiveness
0.11
0.27
0.06
0.42
0.67
CEO
High Trustworthiness
0.46
0.31
0.42
1.46
0.15
Expertise
-0.41
0.32
-0.36 -1.28
0.21
2
Model Statistics
R = 0.20, adjusted R = -0.00, F(3, 64) = 0.92, p = .44
Attractiveness
0.36
0.26
0.24
1.40
0.17
CEO
Low Trustworthiness
-0.05
0.34
-0.05 -0.14
0.89
Expertise
-0.10
0.36
-0.09 -0.27
0.79
Model Statistics

2

R = 0.18, adjusted R = -0.02, F(3, 63) = 0.67, p = .58
Attractiveness
0.30
0.24
1.86
Organization High Trustworthiness
-0.31
0.31
-0.23
Expertise
0.21
0.35
0.15
2
Model Statistics
R = 0.22, adjusted R = 0.00, F(3, 63) = 1.02, p = .39
Attractiveness
0.02
0.23
0.01
Organization Low Trustworthiness
0.05
0.23
0.05
Expertise
0.25
0.25
0.21
Model Statistics

2

R = 0.25, adjusted R = 0.02, F(3, 63) = 1.43, p = .24

46

1.27
-1.01
0.60

0.21
0.32
0.55

0.09
0.22
1.00

0.93
0.82
0.32

DISCUSSION
This study tested the effects of apology crisis responses on attitudes and behavioral
intentions, as a function of source type and virality. This study examined a series of persuasion
effects of source type and virality. The results of this study did not confirm the hypotheses
proposed, but it explored numerous of interesting findings on perceived source credibility.
Findings showed that the main effects of source type and the number of likes and the
interaction between them on attitudes toward the status update, attitudes toward the company,
and viral behavioral intentions, were not significant. There are a number of plausible
explanations for this trend. First, it is possible that the lack of significance was largely influenced
by the choice of crisis response strategy; apology. Past studies (Patel & Reinsch, 2003; Kim,
Avery, &Lariscy, 2009; Claey, Cauberghe, & Vyncke, 2010) have shown that apology strategy
receives the most positive response from participants, which shows the effectiveness of apology
compared with other strategies. It is possible that in this study, apology worked equally well with
different conditions. Future studies should compare different crisis response strategies to see
whether strategy type has influence on subject’s decision-making process in relation to source
type and virality. Second, it is possible that the crisis scenarios, which were crafted to represent
fictitious companies and scenarios, seemed unrealistic for participants, and thus led to the
responses reported in the results section. Future studies could use real life scenarios to increase
the external validity of the study. Additionally, past studies showed weak effects of the exposure
to social media messages and attitudes and viral behavioral intentions (Alhabash, et al., 2013),
which suggests that college students might be cautious in how their evaluations and behavioral
intentions are expressed to social media messages related to companies. It is possible that
participants valued the interpersonal communication with their friends rather than with

47

commercial companies. Moreover, this study was conducted online, subjects may not pay
enough attention during this study, and future study can conduct in a computer lab instead.
The second set of findings related to the effect of perceived source credibility on attitudes
and viral behavioral intentions offers interesting insights to the dynamics of public relations on
social media. First, trustworthiness was the strongest predictor of attitudes toward both the status
update and company in crisis, which was also qualified by similar trends as a function of source
type and number of likes. This means that regardless of the source type or number of likes, the
greater the perceptions of trustworthiness the more effective the crisis response on social media.
This result can also be supported by a lot of studies. Miller and Baseheart (1969)’s study also
showed that message was more persuasive when trustworthiness was high. Another study
(McGinnies et al., 1980) confirmed this finding by comparing trustworthiness and expertise, they
found that messages with trustworthiness always received persuasive effect, while messages
from experts might not.
Another finding showed that the relationship between trustworthiness and attitudes
toward status updates and company was stronger when the CEO or organization status updates
had low number of likes, compared to those with high number of likes. The influence of
trustworthiness in this condition might be driven by lack of peer evaluation, study found that
individual’s performance outcome is effected by peer evaluation (Saavedra & Kwun, 1993), in
this study, when readers receive less validation from others (i.e., they do not have confirmation
of the information), the trustworthiness of source will influence their evaluation of the message.
This study also had some small findings, for example, expertise has been found as
significantly strong predictor to attitude towards status update and company in crisis. It shows
that when participants received messages from an organization with high number of likes,

48

expertise positively predicted people’s evaluation of messages or companies. The authors
speculate that the quality of the brand images (i,e., company’ logos) stimulated readers’
cognitive perception of brand attributions in situations where trustworthiness had been
established by peers (high likes). Even though this study did not use real brands to mitigate
previous bias, the design of company logo can also function as a brand image to persuade
message receivers. Also, in the test about attitude toward the company in crisis, attractiveness
has been shown as a strong predictor when message from CEO with low number of likes. As
expertise is related to organization messages, attractiveness of CEO serves as a factor in
persuasion. As discussed in the literature review, attractiveness is more likely from personal
influence, such as celebrities.
THEORTICAL AND PRACTIAL IMPLICATION
This study investigated attitudes and viral behavioral intention toward companies
undergoing crises. The study provided a reconceptualization of crisis response strategies using
models of persuasion applicable to the new media environment. More specifically, the study
explored the effectiveness of crisis responses shared on popular SNSs as a function of two
message features: source type and virality. The results showed that there were no significant
effects of these two message features. Participants did not feel the difference between message
from the CEO and message from organizations, or other people’s evaluation of the message. The
reasons might be when readers try to evaluate crisis response message with their own cognition
and analysis, such as the response strategy, rather than influence from source type or others’
opinion. This study furthered our understanding of the effects of apology crisis responses within
the context of persuasion. Future studies should explore cognitive effort in relations to other
characteristics of crisis messages. This study has shown that message features has few influences

49

on reader’s attitude, while the vital factor is whether the source readers received is credible or
not. Therefore, this study also helped to lie out the relationship between perceived source
credibility and attitude reaction. Future study can further and build up the framework within
perceived source credibility and other online message features.
On the practical level, this study offered companies and crisis managers with evidencebased tactics to be used in responding to crisis situations. Additionally, the study provided
practitioners with the knowledge pertaining to the effectiveness of social media during crisis
situations, such as the importance of trustworthiness in message delivery, which also implies the
effectiveness of apology strategy in crisis communication.
LIMITATION AND FUTURE STUDY
This study has several limitations, which also inspire future studies. First, this study does
not consider the severity of crisis, even though the pretest eliminated the severity difference
between repetition cases, the result might vary when the crisis severity changes. In the future,
researchers can manipulate the severity of crises with high, medium, low, to explore whether
severity can be another factor influence audiences’ perspective. Second, this study only
manipulated the number of likes. It is possible that subjects did not recognize the different
between messages from CEO and organizations. In the future study, we will check both source
type and number of likes. In the future, researchers should investigate the influence of different
Facebook functions. Third, the subjects of this study were college students, which limits the
generalizability of the results. It is also possible that the issues we chose were not highly relevant
to subjects, which resulted in insignificant result. Future studies should recruit a representative
sample of participants from to enhance external validity. Moreover, this study only applied one
type of crisis response strategy. Future studies should test whether there is a persuasion effect

50

difference between different crisis response strategies. Finally, the crisis scenarios and responses
were all fabricated, which might have harmed the study’s external validity. Future studies should
refine the stimuli construction by selecting real life scenarios and actual responses.

CONCLUSION
This study has discovered reader’s perception when they receive crisis response message
on Facebook. The result showed that message features, such as source type or virality have little
influence on participants’ attitude when they read apology messages. However, this study found
that perceived source credibility (mainly trustworthiness) plays an important role in crisis
communication persuasion. This study emphasizes the importance of credible messages on social
network websites, and provides the foundation of establishing credibility of social media
accounts.

51

APPENDICES

52

Appendix 1. Facebook status update stimuli

Figure 1. CEO response with high likes (case 1)
For interpretation of the references to color in this and all other figures, the reader is referred to
the electronic version of this thesis.

Figure 2. CEO responses with high likes (case 2)

53

Figure 3. CEO response with high likes (case 3)

Figure 4. CEO response with low likes (case 1)

54

Figure 5. CEO response with low likes (case 2)

Figure 6. CEO response with low likes (case 3)

55

Figure 7. Organization response with high likes (case 1)

Figure 8. Organization response with high likes (case 2)

56

Figure 9. Organization response with high likes (case 3)

Figure 10. Organization response with low likes (case 1)

57

Figure 11. Organization response with low likes (case 2)

Figure 12. Organization response with low likes (case 3)

58

Appendix 2. Pre-test Questionnaire
CONSENT FORM
Please read the following form carefully before you start with the experiment. Once you have
read the form and still want to participate, provide consent and continue.
You are invited to participate in a pretest part of a larger study for a Master’s Thesis. When you
are invited to participate in research, you have the right to be informed about the study
procedures so that you can decide whether you want to consent to participation. Please ask the
researcher to explain any words or information that you do not understand.
Your participation in this study is voluntary. You may choose not to participate at all, or you may
refuse to participate in certain procedures or answer certain questions or discontinue your
participation at any time without consequence (e.g., it will not affect treatment you will receive,
will not affect your grade or evaluation, etc.).
Description. This study deals with pretesting experimental stimuli that will be later used in a
larger study. You will be asked to evaluate a number of crisis scenarios, Facebook status updates,
and picture. Finally, you will be asked to report some demographic information.
Risks. Your participation in this study is not expected to cause you any risk greater than those
encountered in everyday life. Your answers will not harm you in any way, nor will they influence
your grades in this course. If you feel any discomfort in answering any question, you can
withdraw from the study without any consequences.
Confidentiality. Your identity, participation, and any information you provide will remain
confidential. This information will not be shared with anyone, and will only be used for the
purpose of the research.
Incentives for Participation. You will receive extra-credit for your participation in this study as
per agreement with your professor/instructor. If you decline to participate in the research study,
you will be given the opportunity to complete a research paper on the role of social media in
society. IN ORDER TO GET CREDIT, MAKE SURE TO ENTER YOUR NAME AND THE
CLASS TO WHICH THE CREDIT WILL COUNT AT THE END OF THE ONLINE
QUESTIONNAIRE. Your name and identity will not be linked in any way to your answers. This
is only to ensure that you get extra or course credit for your participation.
Questions, Concerns and Complaints. If you have any questions about the research, please
contact Shupei Yuan by email: yuanshup@msu.edu or phone: (517) 974-4355.
If you have questions or concerns about your role and rights as a research participant, would like
to obtain information or offer input, or would like to register a complaint about this study, you
may contact, anonymously if you wish, the Michigan State University’s Human Research
Protection Program at 5173552180, Fax 5174324503, or email irb@msu.edu or regular mail at
207 Olds Hall, MSU, East Lansing, MI 48824.
59

I have read this consent form and my questions have been answered. BY CLICKING ON THE
“NEXT’ BUTTON BELOW, I give my consent to participate in this study:
PART ONE: Rating Crisis Scenarios
In this section, you will be exposed to a number of scenarios for crises that faced fictitious
companies that will be later used in a larger study. You can spend as much time as you’d like in
reading each scenario. After reading the scenario, you will be asked to respond to a number of
questions that are briefly described below.
First, we ask you to rate how AROUSED you felt on a 7-point scale, where…
1 = not at all aroused, not at all excited, not at all awake, and
7 = extremely aroused, excited, awake.
Next, we will ask you to rate both how negative and how positive you felt while viewing each
picture. Sometimes the negative rating scale will come first, and sometimes the positive scale
will come first. We want you to rate how negative and positive you felt separately. So you can
feel both negative and positive or just negative or just positive.
You will rate how POSITIVE you felt on a 7-point scale, where…
1 = not at all positive, not at all happy, not at all pleased, and
7 = extremely positive, happy, pleased.
You will also rate how NEGATIVE you felt on a 7-point scale, where…
1 = not at all negative, not at all unhappy, not at all annoyed, and
7 = extremely negative, unhappy, annoyed.
Additionally, we would like you to evaluate the severity of the crisis situation using a 7-point
scale, where…
1 = not at all severe
7 = extremely severe
Case 1:
A plasticizer has been found in the containers of a women’s nutrition supplement named Perfect,
which is produced by company X. The plasticizer is a carcinogenic compound, which has a high
rate of migration into the product. Naturally, this means that pregnant women have been
ingesting a carcinogenic compound for over a year. After the news about this crisis broke,
Company X released an immediate official response. In a press conference, the company’s CEO,
John Smith stated that the company takes full responsibility for this crisis and announced an
immediate recall of all the products from the market.
Does this remind you of a similar crisis case?
 Yes
 No

60

Please rate the scenario… (THIS WILL BE REPEATED AFTER EACH SCENARIO)
Not at all
Extremely
1
2
3
4
5
6
7
AROUSED
AROUSED
Not at all
Extremely
1
2
3
4
5
6
7
NEGATIVE
NEGATIVE
Not at all
Extremely
1
2
3
4
5
6
7
POSITIVE
POSITIVE
Not at all
Extremely
1
2
3
4
5
6
7
SEVERE
SEVERE
Case 2:
Company X is a pharmaceutical company that provides high cost treatments for a crippling
neurological disorder named “Capsleter.” Recently, a customer called the hotline to report that
the labeling looked slightly different than usual. Lab tests determined that the pills were
counterfeit. The Food and Drug Administration (FDA) investigatedthecase and found a large
quantity of counterfeit Capsleter products in the market. Company X, in a press conference
headed by its CEO Steve MacAlisster, claimed full responsibility for this crisis, and issued an
immediate recall for all the products.
Does this remind you of a similar crisis case?
 Yes
 No
Please rate the scenario.
Not at all
1
AROUSED
Not at all
1
NEGATIVE
Not at all
1
POSITIVE
Not at all
1
SEVERE

2

3

4

5

6

7

2

3

4

5

6

7

2

3

4

5

6

7

2

3

4

5

6

7

Extremely
AROUSED
Extremely
NEGATIVE
Extremely
POSITIVE
Extremely
SEVERE

Case 3:
Company X is a soft drink company that manufactures Mango Mania. The company has been
advertising that Mango Mania is produced from 100% mango juice with addedcarbonated water.
It waslater found that the juice is actually composed of many synthetic chemicals thatare far
cheaper thanfreshly-squeezed mangos. After the news broke, the CEO, Donalad MacDougal,
announced that Company X has immediately fired the person responsible for sourcing the
ingredients and apologized to consumers on major media.
Does this remind you of a similar crisis case?
 Yes
 No

61

Please rate the scenario.
Not at all
1
AROUSED
Not at all
1
NEGATIVE
Not at all
1
POSITIVE
Not at all
1
SEVERE

2

3

4

5

6

7

2

3

4

5

6

7

2

3

4

5

6

7

2

3

4

5

6

7

Extremely
AROUSED
Extremely
NEGATIVE
Extremely
POSITIVE
Extremely
SEVERE

Case 4:
Company X is a household appliances company that produces a hairdryer model called
Silkynism. Recently, several women have reported fires resulting from the hair dryers on hot
settings. It was determined that the fire resulted from the wiring in the power cord, but it was
uncertain whether it was an engineering failure or a failure by a supplier company. Company X,
in a press conference by CEO Joe Taylor responded to the public immediately that took full
responsibility for the crisis.
Does this remind you of a similar crisis case?
 Yes
 No
Please rate the scenario.
Not at all
1
AROUSED
Not at all
1
NEGATIVE
Not at all
1
POSITIVE
Not at all
1
SEVERE

2

3

4

5

6

7

2

3

4

5

6

7

2

3

4

5

6

7

2

3

4

5

6

7

Extremely
AROUSED
Extremely
NEGATIVE
Extremely
POSITIVE
Extremely
SEVERE

Case 5:
Company X is a wholesale beef and pork supplier to stores around the United States. A number
of Company X’s consumers have been recently hospitalized within days of eating meat supplied
by this company. Doctors have said that many of the food poisoning victims also showed signs
of toxoplasmosis, which indicates exposure to unclean slaughterhouse. In a press conference held
by CEO Mike Fett, Company X took full responsibility for this crisis, apologized for consumers,
andissued an immediate recall on all products.
Does this remind you of a similar crisis case?
 Yes
 No

62

Please rate the scenario.
Not at all
1
AROUSED
Not at all
1
NEGATIVE
Not at all
1
POSITIVE
Not at all
1
SEVERE

2

3

4

5

6

7

2

3

4

5

6

7

2

3

4

5

6

7

2

3

4

5

6

7

Extremely
AROUSED
Extremely
NEGATIVE
Extremely
POSITIVE
Extremely
SEVERE

Case 6:
A beverage manufacturer Company X makes several “healthy living” products. The Coconut
Water line operates a few hours after the Almond Milk line, the downtime being mainly reserved
for cleaning the equipment. A newspaper reported that several customers with nut allergies were
hospitalized. Investigation showed that this crisis emerged because a new trainee forgot to clean
the equipment during downtime and ran the Coconut Water line.In a press conference held by its
CEO Thomas Clark, Company X apologized for the crisis, fired the employee and took full
responsibility for the crisis.
Does this remind you of a similar crisis case?
 Yes
 No
Please rate the scenario.
Not at all
1
AROUSED
Not at all
1
NEGATIVE
Not at all
1
POSITIVE
Not at all
1
SEVERE

2

3

4

5

6

7

2

3

4

5

6

7

2

3

4

5

6

7

2

3

4

5

6

7

Extremely
AROUSED
Extremely
NEGATIVE
Extremely
POSITIVE
Extremely
SEVERE

Case 7:
Company X is a cosmetic manufacturer of ShinyNail, a well-known nail polish brand. Recently,
several female customers complained that they suffered injuries as a result of using ShinyNail. It
has been found that the damage to the glass vials in unpressurized airplane chambers has caused
shards of glass to be mixed in with the nail polish.In a press conference held by its CEO Jeff
Wilson, Company X apologized and took full responsibility for the crisis, and issued an
immediate recall for ShinyNail.
Does this remind you of a similar crisis case?
 Yes
 No

63

Please rate the scenario.
Not at all
1
AROUSED
Not at all
1
NEGATIVE
Not at all
1
POSITIVE
Not at all
1
SEVERE

2

3

4

5

6

7

2

3

4

5

6

7

2

3

4

5

6

7

2

3

4

5

6

7

Extremely
AROUSED
Extremely
NEGATIVE
Extremely
POSITIVE
Extremely
SEVERE

Case 8
Company X is liquor manufacture that mainly produces whiskey. Recently, the company
received many complaints about an “off flavor,” which, upon further investigation, was the result
of methanol contamination of the product. It was unknown how long methanol had been filled
into the bottles instead of ethanol.In a press conference held by Robert Murphy, Company X
apologize for the crisis and took full responsibility. Additionally, the company issued an
immediate product recall.
Does this remind you of a similar crisis case?
 Yes
 No
Please rate the scenario.
Not at all
1
AROUSED
Not at all
1
NEGATIVE
Not at all
1
POSITIVE
Not at all
1
SEVERE

2

3

4

5

6

7

2

3

4

5

6

7

2

3

4

5

6

7

2

3

4

5

6

7

Extremely
AROUSED
Extremely
NEGATIVE
Extremely
POSITIVE
Extremely
SEVERE

Case 9
Company X has released its first line of full-sized trucks in the United States. After a successful
initial launch, many customers have complained about running out of gas while driving the
automobile. All of the complaints have stemmed from the fuel level being incorrectly reported on
the fuel gauge, leading to mid-drive breakdowns. In a press conference held by its CEO Levine
Bulter, Company X apologized and took full responsibility for this crisis, and issues an
immediate product recall.
Does this remind you of a similar crisis case?
 Yes
 No

64

Please rate the scenario… (THIS WILL BE REPEATED AFTER EACH SCENARIO)
Not at all
Extremely
1
2
3
4
5
6
7
AROUSED
AROUSED
Not at all
Extremely
1
2
3
4
5
6
7
NEGATIVE
NEGATIVE
Not at all
Extremely
1
2
3
4
5
6
7
POSITIVE
POSITIVE
Not at all
Extremely
1
2
3
4
5
6
7
SEVERE
SEVERE
Case 10
Company X is a diaper manufacturer that recently released a new diaper for young infants. The
innovative new absorbent pad in the bottom has been tested to absorb 0.5 L of urine without
leaking. Recently, several mothers complained about large rashes developing after wearing the
diaper. After investigation, it was shown that baby powder has a chemical reaction with the
patented pad. In a press conference held by CEO Robert Morrison, Company X apologized and
took full responsibility for this crisis, and issued an immediate product recall.
Does this remind you of a similar crisis case?
 Yes
 No
Please rate the scenario.
Not at all
1
AROUSED
Not at all
1
NEGATIVE
Not at all
1
POSITIVE
Not at all
1
SEVERE

2

3

4

5

6

7

2

3

4

5

6

7

2

3

4

5

6

7

2

3

4

5

6

7

Extremely
AROUSED
Extremely
NEGATIVE
Extremely
POSITIVE
Extremely
SEVERE

Case 11
Company X is a manufacturer of batteries. After several years of making car batteries, they
decided to producethe AA and AAA battery models. Shortly after the launch, it has been reported
that batteries in several expensive appliances are leaking fluid and causing hazards in homes
across the United States. In a press conference held by its CEO, George Hughes, Company X
apologized and took full responsibility for the crisis, and issued an immediate recall.
Does this remind you of a similar crisis case?
 Yes
 No

65

Please rate the scenario.
Not at all
1
AROUSED
Not at all
1
NEGATIVE
Not at all
1
POSITIVE
Not at all
1
SEVERE

2

3

4

5

6

7

2

3

4

5

6

7

2

3

4

5

6

7

2

3

4

5

6

7

Extremely
AROUSED
Extremely
NEGATIVE
Extremely
POSITIVE
Extremely
SEVERE

Case 12
Company X is a video game console manufacturer that released a new console. Recently, a large
number of customers complained that, after a couple months, the product simply stops working.
The causes of this malfunction are yet to be determined by an internal investigation. As of yet,
the only possible fix is to send the gaming console back to Company X to troubleshoot the rootcause analysis.In a press conference held by Derek Sanders, Company X apologized and took
full responsibility for this crisis, and encouraged consumers to send their consoles for
troubleshooting.
Does this remind you of a similar crisis case?
 Yes
 No
Please rate the scenario.
Not at all
1
AROUSED
Not at all
1
NEGATIVE
Not at all
1
POSITIVE
Not at all
1
SEVERE

2

3

4

5

6

7

2

3

4

5

6

7

2

3

4

5

6

7

2

3

4

5

6

7

Extremely
AROUSED
Extremely
NEGATIVE
Extremely
POSITIVE
Extremely
SEVERE

PART TWO: Evaluating Pictures
Next, you are going to rate different pictures. We’d like you to rate each picture using the
following scales:
ATTRACTIVENESS
1 = not at all attractive
7 = extremely attractive
TRUSTWORTHNESS
1 = not at all trustworthy
7 = extremely trustworthy
EXPERTISE
1 = not at all experienced
2 = extremely experienced
66

Please rate the picture… (THIS WILL BE REPEATED AFTER EACH PICTURE)
Not at all
Extremely
1
2
3
4
5
6
7
ATTRACTIVE
ATTRACTIVE
Not at all
1
2
3
4
5
6
7
Extremely
TRUSTWORTHY
TRUSTWORTHY
Not at all
1
2
3
4
5
6
7
Extremely
EXPERIENCED
EXPERIENCED

No. 1

No.2

No.3

No.4

No.5

No.6

No.7

No.8

No.9

No.10

No.11

No.12

Figure 13. CEO Images used in study
67

Figure
PART THREE: Evaluating Logos
Next, you are going to rate different company logos. We’d like you to rate each picture using the
following scales:
ATTRACTIVENESS
1 = not at all attractive
7 = extremely attractive
TRUSTWORTHNESS
1 = not at all trustful
7 = extremely trustful
EXPERTISE
1 = not at all experienced
2 = extremely experienced

Please rate the logo… (THIS WILL BE REPEATED AFTER EACH LOGO)
Not at all
1
2
3
4
5
6
7
ATTRACTIVE
Not at all
1
2
3
4
5
6
7
TRUSTWORTHY
Not at all
1
2
3
4
5
6
7
EXPERIENCED

68

Extremely
ATTRACTIVE
Extremely
TRUSTWORTHY
Extremely
EXPERIENCED

No. 1

No.2

No.3

No.4

No.5

No.6

No.7

No.8

No.9

No.10

No.11

No.12

Figure 14. Company Logos
PART FOUR: Demographic Information
What is your gender?
 Female
 Male
 Transgender
In what year were you born? [drop-down menu]
Which of the following best describes your current relationship situation?






Single
Dating, but it’s not committed
Dating and it’s a committed relationship
Living with someone in a romantic relationship
Engaged but not living together
69

 Engaged and living together
 Married
 Other, please specify: ___________________________________
What is your current class standing?








Freshman
Sophomore
Junior
Senior
MA
Ph.D.
Other, please specify

Which of the following best describes your ethnicity?








American Indian or Alaska Native
Asian
Black or African American
Hispanic or Latino
Native Hawaiian or Other Pacific Islander
White
Other, please specify

What is your total annual family income?
 Less than $10,000
 $10,000 to $24,999
 $25,000 to $49,999
 $50,000 to $74,999
 $75,000 to $99,999
 $100,000 to $124,999
 $125,000 to $149,999
 $150,000 or more

70

Appendix 3. Main test questionnaire
CONSENT FORM
Please read the following form carefully before you start with the experiment. Once you have
read the form and still want to participate, provide consent and continue.
You are invited to participate in an experiment for a Master’s Thesis. When you are invited to
participate in research, you have the right to be informed about the study procedures so that you
can decide whether you want to consent to participation. Please ask the researcher to explain any
words or information that you do not understand.
Your participation in this study is voluntary. You may choose not to participate at all, or you may
refuse to participate in certain procedures or answer certain questions or discontinue your
participation at any time without consequence (e.g., it will not affect treatment you will receive,
will not affect your grade or evaluation, etc.).
Description. This study deals evaluating crisis situations and responses. First, you will answer a
few questions about your media and Facebook use. Second, you will read three crisis scenarios
followed by crisis responses posted on Facebook, where you will be asked to evaluate them.
Finally, you will be asked to report some demographic information.
Risks. Your participation in this study is not expected to cause you any risk greater than those
encountered in everyday life. Your answers will not harm you in any way, nor will they influence
your grades in this course. If you feel any discomfort in answering any question, you can
withdraw from the study without any consequences.
Confidentiality. Your identity, participation, and any information you provide will remain
confidential. This information will not be shared with anyone, and will only be used for the
purpose of the research.
Incentives for Participation. You will receive extra-credit for your participation in this study as
per agreement with your professor/instructor. If you decline to participate in the research study,
you will be given the opportunity to complete a research paper on the role of social media in
society. IN ORDER TO GET CREDIT, MAKE SURE TO ENTER YOUR NAME AND THE
CLASS TO WHICH THE CREDIT WILL COUNT AT THE END OF THE ONLINE
QUESTIONNAIRE. Your name and identity will not be linked in any way to your answers. This
is only to ensure that you get extra or course credit for your participation.
Questions, Concerns and Complaints. If you have any questions about the research, please
contact Shupei Yuan by email: yuanshup@msu.edu or phone: (517) 974-4355.
If you have questions or concerns about your role and rights as a research participant, would like
to obtain information or offer input, or would like to register a complaint about this study, you
may contact, anonymously if you wish, the Michigan State University’s Human Research
Protection Program at 5173552180, Fax 5174324503, or email irb@msu.edu or regular mail at
71

207 Olds Hall, MSU, East Lansing, MI 48824.
I have read this consent form and my questions have been answered. BY CLICKING ON THE
“NEXT’ BUTTON BELOW, I give my consent to participate in this study:
INTRODUCTION
A crisis is any expected that might lead to, an unstable and dangerous situation affecting an
individual or, group, community or whole society. After crisis happened, as a company or
organization, there are many different ways to respond, such as deny this crisis, find a scapegoat,
show compensation or apologize. You will be asked a series of question about your
understanding of crisis response and your online behavior.
PART ONE: FACEBOOK USE HABITS
Do you have a Facebook account?
 YES
 NO
When did you join Facebook?[drop-down menu]
In the past week, on average, approximately how many minutes PER DAY have you spent
actively using Facebook?[drop-down menu]
Approximately, how many commercial companies and brands do you interact with on Facebook
(i.e., liked their Fan page, added the company/brand as a friend, etc.)[drop-down menu]
The following statements deal with how you feel about Facebook. Indicate your
agreement/disagreement with each of the statements using the scale from 1=Strongly Disagree to
7=Strongly Agree.
Strongly
Disagree
1

Strongly
Agree
2

3

4

5

6

7

Facebook is part of my everyday activity


      

I am proud to tell people I'm on Facebook


      

Facebook has become part of my daily routine


      

I feel out of touch when I haven't logged onto Facebook
for a while

      


 I feel I am part of the Facebook community


      

I would be sorry if Facebook shut down

      

The following statements deal with specific features the people usually use on Facebook.
Indicate your agreement/disagreement with each of the statements using the scale from
1=Strongly Disagree to 7=Strongly Agree.
72

Strongly
Disagree
1

Strongly
Agree
2

3

4

5

6

7

I update my status on Facebook often

      

I use the comments feature on Facebook often.

      

I write Wall posts on my friends’ pages often.

      

I use private messages feature on Facebook often.

      

I use Facebook chat often.

      

I use Facebook Groups often.

      

I use applications often.

      

I use the ‘Causes’ feature often.

      

I play games on Facebook often.

      

The following statements specifically deal with how you interact with commercial companies
and/or brands on Facebook. Indicate your agreement/disagreement with each of the statements
using the scale from 1=Strongly Disagree to 9=Strongly Agree.
Strongly
Disagree
1

Strongly
Agree
2

3

4

5

6

7

I join Facebook groups for companies/brands often.

      

I “like” fan pages of companies/brands often.

      

I add employees of companies/brands as friends on
Facebook often.

      

I write Wall posts on company/brand’s pages often.

      

I “like” status updates posted by companies/brands often.

      

I “share” status updates posted by companies/brands often.

      

I “comment” on status updates posted by
companies/brands often.

      

I “like” pictures posted by companies/brands often.

      

I “share” pictures posted by companies/brands often.

      

I “comment” on pictures posted by companies/brands
often.

      

I “like” videos posted by companies/brands often.

      

I “share” videos posted by companies/brands often.

      

73

I “comment” on videos posted by companies/brands often.

      

The following section deals with the reasons you have for using Facebook. Indicate your
agreement or disagreement with each statement using the following 9-point scale: 1=Strongly
Disagree, and 9=Strongly Agree.
Strongly
Disagree
1

Strongly
Agree
2

3

4

5

6

7

I use Facebook to share information.

      

I use Facebook to share information useful to people.

      

I use Facebook to present information on my interests.

      

I use Facebook to record what I do in life.

      

I use Facebook to record what I have learned.

      

I use Facebook to record where I have been.

      

I use Facebook to connect with people who share some
of my values.

      

I use Facebook to connect with people who are similar to
me.

      

I use Facebook to meet new people.

      

I use Facebook because it is enjoyable.

      

I use Facebook because it entertains me.

      

I use Facebook because it helps me pass the time.

      

I use Facebook because I have nothing better to do.

      

I use Facebook because it relaxes me.

      

I use Facebook to show my personality.

      

I use Facebook to tell others about myself.

      

I use Facebook because I like that I can post things I
want to say immediately.

      

I use Facebook because it is easy to use.

      

I use Facebook because it is convenient.

      

74

Patrick Collins Inc. is a manufacturer of batteries. After several years of making car batteries,
they decided to produce the AA and AAA battery models. Shortly after the launch, it has been
reported that batteries in several expensive appliances are leaking fluid and causing hazards in
homes across the United States. In a press conference held by its CEO, George Hughes, the
company apologized and took full responsibility for the crisis, and issued an immediate product
recall.

Figure 15. Facebook Status Test Material Example
Other stimuli cases:
A beverage manufacturer Steven Adams makes several “healthy living” products. The Coconut
Water line operates a few hours after the Almond Milk line, the downtime being mainly reserved
for cleaning the equipment. A newspaper reported that several customers with nut allergies were
hospitalized. Investigation showed that this crisis emerged because a new trainee forgot to clean
the equipment during downtime and ran the Coconut Water line. In a press conference held by its
CEO Thomas Clark, the company apologized for the crisis, fired the employee and took full
responsibility for the crisis.
Henry Reed L.L.C is liquor manufacture that mainly produces whiskey. Recently, the company
received many complaints about an “off flavor”, which, upon further investigation, was the result
of methanol contamination of the product. It was unknown how long methanol had been filled
into the bottles instead of ethanol. In a press conference held by its CEO, Robert Murphy, the
company apologized for the crisis and took full responsibility. Additionally, the company issued
an immediate product recall.
Next, you will see the screenshot of the company’s response on Facebook:
[QUESTIONS TO BE ASKED AFTER EACH CASE]
Please evaluate the specific Facebook crisis response message.
75

POSITIVE















NEGATIVE

BAD















GOOD

UNFAVORABLE















FAVORABLE

Please evaluate the company “Patrick Collines”
POSITIVE















NEGATIVE

BAD















GOOD

UNFAVORABLE















FAVORABLE

Even if you don’t have a Facebook account, please indicate whether you would pass the message
along or not.
Strongly
Disagree
1

Strongly
Agree
2

3

4

5

6

7

This response message is worth sharing with others

      

I would recommend this response message to others

      

I would “Like” this response message

      

I would “Share” this response message

      

I would “Comment” on the response message

      

I would ignore this response message

      

How do you evaluate this company?
Dependable















Not Dependable

Honest















Dishonest

Reliable















Unreliable

Sincere















Insincere

Trustworthy















Untrustworthy

Expert















Not Expert

Experienced















Inexperienced

Knowledgeable















Unknowledgeable

Qualified















Not qualified

Skilled















Unskilled

76

Attractive















Unattractive

Classy















Not classy

Beautiful















Ugly

Elegant















Plain

Sexy















Not sexy

Do you remember the number of likes in the message screenshot you just saw?
 High likes
 Low likes
PART THREE: DEMOGRAPHIC INFORMATION
What is your gender?
 Female
 Male
 Transgender
In what year were you born? [drop-down menu]
Which of the following best describes your current relationship situation?









Single
Dating, but it’s not committed
Dating and it’s a committed relationship
Living with someone in a romantic relationship
Engaged but not living together
Engaged and living together
Married
Other, please specify: ___________________________________

What is your current class standing?








Freshman
Sophomore
Junior
Senior
MA
Ph.D.
Other, please specify

Which of the following best describes your ethnicity?
 American Indian or Alaska Native
 Asian
77







Black or African American
Hispanic or Latino
Native Hawaiian or Other Pacific Islander
White
Other, please specify

What is your total annual family income?
 Less than $10,000
 $10,000 to $24,999
 $25,000 to $49,999
 $50,000 to $74,999
 $75,000 to $99,999
 $100,000 to $124,999
 $125,000 to $149,999
 $150,000 or more

78

FOOTNOTE
1

The subjects recruited from SONA system had 51% female (Mean = 1.51, SD = 0.50),

subjects recruited from classes had 69% female (Mean = 1.69, SD = 0.49); subjects’ average age
from SONA system is 21 (Mean = 20.86, SD = 1.70), subjects’ average age from classes
recruiting is 22 (Mean = 21.87, SD = 1.40); the majority of subjects from SONA system were
Caucasian (60.99%), followed by Asian (19.15%), African American (14.18%) and others
(5.67%), the majority of subjects from classes recruiting is Caucasian as well (79.69%), followed
by Asian (12.50), African American (5.47%), and others (2.30%).

79

REFERENCES

80

REFERENCES

Ajzen, I. (1985). From intention to actions: A theory of planned behavior. In J. Kuhl& J.
Beckmann (Eds.), Action-control: From cognition to behavior (pp.11-39). Heidelberg:
Srpinger.
Ajzen, I. (1991). The theory of planned behavior. Organizational behavior and human decision
processes, 50(2), 179-211.
Fishbein, M., &Ajzen, I. (1980).Understanding attitudes and predicting social behavior. Prendice
Hall. New York.
Alhabash, S., McAlister, R. A., Rifon, J. N., Quilliam, T. E., Sternadori, M., Richards, I. J.,
(2013b). Redefining Virality in Less Broad Strokes: Predicting Viral Behavioral Intentions
from Motivations and Uses of Facebook and Twitter. American Academy of Advertising
Annual Conference.Alberquerque, NM.
Alhabash, S., McAlister, R. A., Hagerstrom, A., Quilliam, T. E., Rifon, J. N., Richards, I. J.,
(2013a). Between Likes and Shares: Effects of Emotional Appeal and Virality on the
Persuasiveness of Anticyberbullying Messages on Facebook. Cyberpsychology, Behavior,
and Social network.16, 1-7.
American Red Cross (2011).Social Media in Disasters and Emergencies. Retrieved from
http://www.redcross.org/portal/site/en/menuitem.94aae335470e233f6cf911df43181aa0/?vg
nextoid=7a82d1efe68f13 10VgnVCM10000089f0870aRCRD
Arndt, J. (1967). Word-of-mouth advertising and informal communication.In D. Cox (Ed.), Risk
taking and information handling in consumer behaviour. Boston: Harvard University
Austin, L., Liu, B. F., & Jin, Y. (2012). How audiences seek out crisis information: Exploring the
social-mediated crisis communication model. Journal of Applied Communication Research,
40(2), 188-207.
Arndt, J. (1967). Word of mouth advertising and informal communication. Risk taking and
information handling in consumer behavior, 188-239.
Barnes, J. G. (1978). A hierarchical model of source effects in retail newspaper
advertising.Advances in Consumer Research, 5(1), 235–242.
Barnes, N., & Mattson, E. (2008, October). Social media in the inc. 500: The first longitudinal
EBSCOhost.study. Journal of New Communications Research, 3(1), 74-78. Retrieved
February 11.
Bonchi, F., Castillo, C., &Ienco, D. (2011).Meme ranking to maximize posts virality in
microblogging platforms. Journal of Intelligent Information Systems, 1-29.
81

Benoit, W. L. (1997). Image Repair Discourse and Crisis Communication. Public Relations
Review 23, 177-186.
Benoit, W. L., & Drew, S. (1997). Appropriateness and effectiveness of image repair strategies.
Communication Reports, 10, 153–163.
Borkenguitat song gets airline's attention. (2009, July 8) CBC News. Retrieved from
http://www.cbc.ca/news/arts/music/story/2009/07/08/united-breaks-guitars.html
boyd, D. M., & Ellison, N. B. (2007). Social Network Sites: Definition, History, and Scholarship.
Journal of Computer-Mediated Communication, 13(1), 210-230.
Bradford, J. L., & Garrett, D. E. (1995). The effectiveness of corporate communicative responses
to accusations of unethical behavior.Journal of Business Ethics,14, 875–892.
Brenner, J. (2013, Feb 5). Pew Internet: Social networking (full detail). Pew Internet & American
Life Project. Retrieved from http://pewinternet.org/Commentary/2012/March/Pew-InternetSocial-Networking-full-detail.aspx.
Bush, A. J., Moncrief, W. C., &Zeithaml, V. A. (1987). Source effects in professional services
advertising. Current Issues & Research in Advertising, 10(1), 153–171.
Buttny, R. (1993). Social accountability in communication. Sage Publications Limited.
Cacioppo, J. T., Petty, R. E., Kao, C. F., & Rodriguez, R. (1986). Central and peripheral routes to
persuasion: An individual difference perspective. Journal of personality and social
psychology, 51(5), 1032.
Cancel, A. E., Cameron, G. T., Sallot, L. M., &Mitrook, M. A. (1997). It depends: a contingency
theory of accommodation in public relations.Journal of Public Relations Research, 9(1), 3163
CEO Survey (2012). Inc.com. Retrieved from http://www.inc.com/magazine/201209/incstaff/survey-of-inc-500-ceos-who-they-are.html
Chaiken, S. (1980). Heuristic versus systematic information processing and the use of source
versus message cues in persuasion. Journal of Personality and Social Psychology, 39(5),
752-766.
Chu, S. C. (2011). Viral advertising in social media: participation in Facebook Groups and
Responses among college-Aged users. Journal of Interactive Advertising, 12(1), 30-43.
Claeys, A.S., Cauberghe, V., &Vyncke, P. (2010). Restoring reputations in times of crisis: An
experimental study of the Situational Crisis Communication Theory and the moderating
effects of locus of control. Public Relations Review,36(3), 256-262. Elsevier Inc.
Coombs, W. T. (1995). Choosing the Right Words The Development of Guidelines for the
Selection of the “Appropriate” Crisis-Response Strategies. Management Communication
82

Quarterly, 8(4), 447-476.
Coombs, W. T. (1999).Ongoing Crisis Communication: Planning, Managing and Responding.
Thousand Oaks, CA: Sage.
Coombs, W. T. (2007a). Protecting organization reputations during a crisis: The development and
application of situational crisis communication theory. Corporate Reputation Review, 10(3),
163-176.
Coombs, W. T. (2007b).Attribution theory as a guide for post-crisis communication
research.Public Relations Review, 33(2), 135-139.
Coombs, W.T. and Holladay, S.J. (1996) 'Communication and attributions in a crisis: An
experimental study of crisis communication', Journal of Public Relations Research, 8, 279–
295.
Coombs, W. T., & Holladay, S. J. (2005). An exploratory study of stakeholder emotions: Affect
and crises. Research on emotion in organizations, 1, 263-280.
Cosh, C. (n.d.) A man and his guitar. Financial Post. Retrieved from
http://www.financialpost.com/scripts/story.html?id=f9065720-c55a-4612-84ebe168fd37ed1f&k=15437.
Craig, S. C., & McCann, J. M. (1978). Assessing communication effects of energy conservation.
Journal of Consumer Research, 5, 82–88.
Dean, D. H. (2004).Consumer reaction to negative publicity effects of corporate reputation,
response, and responsibility for a crisis event. Journal of Business Communication, 41(2),
192-211.
Dekay, S. H. (2012). How large companies react to negative facebook comments. Corporate
Communications, 17(3), 289-299.
Duan, W., Gu, B., &Whinston, A. B. (2008). Do online reviews matter?—An empirical
investigation of panel data. Decision Support Systems, 45(4), 1007-1016.
Dugan, L. (2012, May 15) Facebook, Twitter, Google+, Pinterest: The users of social media.
Media Bistro. Retrieved from http://www.mediabistro.com/alltwitter/social-mediausers_b22556.
Eagly, A. H., &Chaiken, S. (1993). The psychology of attitudes.Harcourt Brace Jovanovich
College Publishers.
ExactTarget releases new research on 'the Meaning of like' with facebook. (2011). Professional
Services Close - Up, Retrieved from
http://ezproxy.msu.edu/login?url=http://search.proquest.com/docview/893964689?accountid
=12598

83

Fazio, R. H. (1990). The MODE model as an integrative framework. Advances in experimental
social psychology, 23, 75-109.
Felix, S.(2012, Aug 17). The 20 Most-Liked Facebook Companies Ever. Business Insider.
Retrieved from http://www.businessinsider.com/the-20-most-liked-companies-onfacebook-2012-8?op=1.
Fuchs-Burnett, T. (2002). Mass public corporate apology. Dispute Resolution Journal, 57(3), 26–
32.
Goldberg, M. E., &Hartwick, J. (1990). The effects of advertiser reputation and extremity of
advertising claim on advertising effectiveness. Journal of Consumer Research, 17, 172–179.
Greenfield, D. (2009, July 13). United Airlines Online Public Response to Dave Carroll YouTube
Video: 9 Tweets. Socialmedia Today. Retrieved from
http://socialmediatoday.com/SMC/109126.
Grunig, J. E. (1992). Communication, public relations, and effective organizations: An overview
of the book. Excellence in public relations and communication management, 1-28.
Grunig, J.E., & Hunt, T. (1984) Managing Public Relations. Fort Worth, TX: Harcourt Brace.
Guerini, M., Pepe, A., &Lepri, B. (2012). Do linguistic style and readability of scientific
abstracts affect their virality. Proceedings of ICWSM.
Hendrick, J. (2011). Managing through a business crisis with social media. Defined Logic Blog.
Retrieved from http://www.definedlogic.com/2011/11/managing-through-a-business-crisiswith-social-media/.
Hochschild, A. R. (1979). Emotion work, feeling rules, and social structure. American Journal of
Sociology, 85, 551-575.
Horton, J. L. (2009). Are corporations using Facebook? Retrieved from http://docs.google.com/
viewer?a=v&q=cache:HjnJqRejSYwJ:www.online-pr.com/Holding/Are%252520Corps%
252520Using%252520Facebook.pdf+number+of+corporations+on+Facebook&hl=en&gl=u
s&pid
=bl&srcid=ADGEESgJp1qECAC3pvZkt2B0RFQQW1ZNFmephoya8kNknqy8cn59f5AXf
qVb5Y5enMZ NaAY7JhRpkw9AycdBqH7q7cNj8iSBKP9hkrg4Km1ZU5cDLaw9Ic_e3JQIHK1puvXkKa5vueE&sig=AHIEtbRe3zaZdACvQL4l8ZM7VUXKTLq5ag
Hovland, C. I., Janis, I. L., & Kelley, H. H. (1963). Communication and persuasion:
psychological studies of opinion change (Vol. 76). Yale University Press.
Jansen, B. J., Zhang, M., Sobel, K., &Chowdury, A. (2009). Twitter power: Tweets as electronic
word of mouth. Journal of the American Society for Information Science and Technology,
60(11), 2169.
84

Jin, C. H., & Yeo, H. C. (2011). Satisfaction, corporate credibility, CEO reputation and
leadership effects on public relationships.Journal of Targeting, Measurement and Analysis
for Marketing, 19(2), 127-140.
Jones, E. E., &McGillis, D. (1976). Correspondent inferences and the attribution cube: A
comparative reappraisal. In J. H. Harvey, W. J. Ickes, & R. F. Kidd (Eds.), New Directions
in Attribution Research (Vol. 1, 389-420). Hillsdale, NJ: Erlbaum.
Kaplan, A. M., &Haenlein, M. (2010). Users of the world, unite! The challenges and
opportunities of Social Media.Business horizons, 53(1), 59-68.
Kim, S., Avery, E. J., &Lariscy, R. W. (2009). Are crisis communicators practicing what we
preach?: An evaluation of crisis response strategy analyzed in public relations research from
1991 to 2009. Public Relations Review, 35(4), 446-448.
Klikauer, T. (2012). A general motors works council's response to the capitalist global financial
crisis: A case study from Germany. Capital and Class, 36(2), 303-322.
Knobel, M., &Lankshear, C. (2008).Digital literacy and participation in online social networking
spaces.Digital Literacies: Concepts, Policies and Practices. New York: Peter Lang, 249-78.
Krebs, J. R., & Dawkins, R. (1984). Animal signals: mind-reading and
manipulation. Behavioural Ecology: an evolutionary approach, 2, 380-402.
Lafferty, B. A., Goldsmith, R. E., & Newell, S. J. (2002). The dual credibility model: The
influence of corporate and endorser credibility on attitudes and purchase intentions. Journal
of Marketing Theory and Practice, 10(3), 1–12.
Lafferty, B. A., Goldsmith, R. E., & Newell, S. J. (2002). The dual credibility model: The
influence of corporate and endorser credibility on attitudes and purchase intentions. Journal
of Marketing Theory and Practice, 10(3), 1–12.
Lalonde, C. (2007), The Potential Contribution of the Field of Organizational Development to
Crisis Management. Journal of Contingencies and Crisis Management, 15: 95–104.
Lazare, A. (1995, Go ahead say you're sorry. Psychology Today, 28, 40-43+.
Lenhart, A., Purcell, K., Smith, A., Zickuhr, K. (2010, Feb 3). Social media and young
aduits.Pew Internet & American Life Project. Retrieved from
http://www.pewinternet.org/Reports/2010/Social-Media-and-Young-Adults.aspx.
MacKenzie, S. B., Lutz, R. J., & Belch, G. E. (1986). The role of attitude toward the ad as a
mediator of advertising effectiveness: A test of competing explanations. Journal of
marketing research, 130-143.
Maenaka, M. (2007). A case study: A crisis management of a chemical company. (inJapanese.
with English summary.). Osaka Economic Papers, 57(2), 70-82.

85

McGuire, W. J. (1985). Attitudes and attitude change. Handbook of social psychology, 2(3), 233346.
Miller, G. R., & Baseheart, J. (1969). Source trustworthiness, opinionated statements, and
response to persuasive communication.
McGinnies, E., & Ward, C. D. (1980). Better Liked Than Right Trustworthiness and Expertise as
Factors in Credibility. Personality and Social Psychology Bulletin, 6(3), 467-472.
Miller, K. (2002). Communication theories. McGraw-Hill.
Miriam, J. M., Andrew, J. F., & Ryan, B. M. (2010). Social and heuristic approaches to
credibility evaluation online.Journal of Communication, 60(3), 413.
Naveed, N. (2012). Role of social media on public relation, brand involvement and brand
commitment.Interdisciplinary Journal of Contemporary Research in Business, 3(9), 904913.
Nelson-Field, K., Riebe, E., & Sharp, B. (2012). What's Not to" Like?" Can a Facebook Fan
Base Give a Brand the Advertising Reach it Needs? Journal of Advertising Research, 52(2),
262.
Ohanian, R. (1990). Construction and validation of a scale to measure celebrity endorsers'
perceived expertise, trustworthiness, and attractiveness. Journal of Advertising, 39-52.
Olson, J. M., Roese, N. J., and Zanna, M. P. (1996). Expectancies. In Higgins, E. T., and
Kruglanski, A. W. (Eds.), Social Psychology: Handbook of Basic Principles (211–238).
New York: Guilford.
Park, H., & Lee, H. (2013). Show Us You Are Real: The Effect of Human-Versus-Organizational
Presence on Online Relationship Building Through Social Networking Sites.
Cyberpsychology, Behavior, and Social Networking.
Patel, A., &Reinsch, L. (2003). Companies can apologize: Corporate apologies and legal
liability. Business Communication Quarterly, 66(1), 9-25.
Petty, R. E., &Cacioppo, J. T. (1990). Involvement and persuasion: Tradition versus integration.
Psychological Bulletin,107(3), 367-374.
Petty, R. E., & Wegener, D. T. (1999). The elaboration likelihood model: Current status and
controversies. New York, NY, US: Guilford Press, New York, NY.
Petty, R. E., Brinõl P., &Priester, J. R. (2008). Implications of the Elaboration Likelihood Model
of Persuasion.Media effects: Advances in theory and research, 10, 125.
Petty, R. E., Cacioppo, J. T., & Schumann, D. (1983). Central and peripheral routes to
advertising effectiveness: The moderating role of involvement.Journal of consumer
research, 135-146.

86

Rudaina, O. Y. (2012). The extent of facebook users' interest in the advertising
messages.International Journal of Marketing Studies, 4(3), 122-133.
Saavedra, R., & Kwun, S. K. (1993). Peer evaluation in self-managing work groups. Journal of
Applied Psychology, 78(3), 450.

Schweiger, W. (2000). Media credibility—experience or image? A survey on the credibility of
the World Wide Web in Germany in comparison to other media.European Journal of
Communication, 15, 37–59.
Sheppard, B. H., Hartwick, J., &Warshaw, P. R. (1988). The theory of reasoned action: A metaanalysis of past research with recommendations for modifications and future research.
Journal of Consumer research, 325-343.
Skeels, M. M., &Grudin, J. (2009, May). When social networks cross boundaries: a case study of
workplace use of facebook and linkedin. InProceedings of the ACM 2009 international
conference on supporting group work.ACM, 95-104.
Sledgianowski, D. &Kulviwat, S. (2009). Using Social Network Sites: The Effects of Playfulness,
Critical Mass and Trust in a Hedonic Context. Journal of Computer Information Systems,
49,(4), 74-83.
Sterne, J. (2010). Social Media Metrics: How to Measure and Optimize Your marketing
investment. Wiley, 272.
Sundar, S. S., &Nass, C. (2001). Conceptualizing sources in online news.Journal of
Communication, 51, 52–72.
Tucker, C. (2011). Virality, Network Effects and Advertising Effectiveness.NET Institute
Working Papers.
United Breaks Guitars (2011, April 29) Fox News. Retrieved from
http://video.foxnews.com/v/3935170/united-breaks-guitars
Vejacka, M. (2012). Facebook advertising and its efficiency on the Slovak
market.E+MEkonomie a Management, (1), 116-127.
Waters, R. D., Burnett, E., Lamm, A., & Lucas, J. (2009). Engaging stakeholders through social
networking: How nonprofit organizations are using Facebook. Public Relations Review, 35,
102–106.
Weiss, W. (1957). Opinion congruence with a negative source on one issue as a factor
influencing agreement on another issue.Journal of Abnormal and Social Psychology, 54(2),
180-186.
Wilson, E. J., &Sherrell, D. L. (1993). Source effects in communication and persuasion research:
87

A meta-analysis of effect size. Journal of the Academy of Marketing Science, 21(2), 101112.
Wilson, T. D., Lisle, D. J., Kraft, D., & Wetzel, C. G. (1989). Preferences as expectation-driven
inferences: effects of affective expectations on affective experience. Journal of Personality
and Social Psychology, 56(4), 519.
Zanna, M. P., &Rempel, JK (1988). Attitudes: A new look at an old concept.The social
psychology of knowledge, 315-334.

88