Management Communication Quarterly
The online version of this article can be found at:
2002 16: 165Management Communication Quarterly
W. Timothy Coombs and Sherry J. Holladay
Helping Crisis Managers Protect Reputational Assets: Initial Tests of the Situational Crisis Communication
can be found at:Management Communication QuarterlyAdditional services and information for
What is This?
- Nov 1, 2002Version of Record >>
at LUND UNIVERSITY LIBRARIES on November 6, 2013mcq.sagepub.comDownloaded from at LUND UNIVERSITY LIBRARIES on November 6, 2013mcq.sagepub.comDownloaded from at LUND UNIVERSITY LIBRARIES on November 6, 2013mcq.sagepub.comDownloaded from at LUND UNIVERSITY LIBRARIES on November 6, 2013mcq.sagepub.comDownloaded from at LUND UNIVERSITY LIBRARIES on November 6, 2013mcq.sagepub.comDownloaded from at LUND UNIVERSITY LIBRARIES on November 6, 2013mcq.sagepub.comDownloaded from at LUND UNIVERSITY LIBRARIES on November 6, 2013mcq.sagepub.comDownloaded from at LUND UNIVERSITY LIBRARIES on November 6, 2013mcq.sagepub.comDownloaded from at LUND UNIVERSITY LIBRARIES on November 6, 2013mcq.sagepub.comDownloaded from at LUND UNIVERSITY LIBRARIES on November 6, 2013mcq.sagepub.comDownloaded from at LUND UNIVERSITY LIBRARIES on November 6, 2013mcq.sagepub.comDownloaded from at LUND UNIVERSITY LIBRARIES on November 6, 2013mcq.sagepub.comDownloaded from at LUND UNIVERSITY LIBRARIES on November 6, 2013mcq.sagepub.comDownloaded from at LUND UNIVERSITY LIBRARIES on November 6, 2013mcq.sagepub.comDownloaded from at LUND UNIVERSITY LIBRARIES on November 6, 2013mcq.sagepub.comDownloaded from at LUND UNIVERSITY LIBRARIES on November 6, 2013mcq.sagepub.comDownloaded from at LUND UNIVERSITY LIBRARIES on November 6, 2013mcq.sagepub.comDownloaded from at LUND UNIVERSITY LIBRARIES on November 6, 2013mcq.sagepub.comDownloaded from at LUND UNIVERSITY LIBRARIES on November 6, 2013mcq.sagepub.comDownloaded from at LUND UNIVERSITY LIBRARIES on November 6, 2013mcq.sagepub.comDownloaded from at LUND UNIVERSITY LIBRARIES on November 6, 2013mcq.sagepub.comDownloaded from at LUND UNIVERSITY LIBRARIES on November 6, 2013mcq.sagepub.comDownloaded from at LUND UNIVERSITY LIBRARIES on November 6, 2013mcq.sagepub.comDownloaded from
10.1177/089331802237233MANAGEMENT COMMUNICATION QUARTERLY /NOVEMBER 2002Coombs, Holladay / INITIAL TESTS
Initial Tests of the Situational Crisis
W. TIMOTHY COOMBS
SHERRY J. HOLLADAY
Eastern Illinois University
AUTHORS’NOTE: An earlier version of this article was pre
sented at the International Communication Association
Convention, Washington, D.C., May 2001.
theory . . .
strategies to the
Management Communication Quarterly, Vol. 16, No. 2, November 2002 165-186
© 2002 Sage Publications
A situational crisis communication theory (SCCT), which articu
lates the variables, assumptions, and relationships that should be
considered in selecting crisis response strategies to protect an orga
nization’s reputation, is advanced. Although various studies taking
a situational approach have touched on certain of the theory’s vari
ables and relationships, this study represents the first attempt to
articulate and begin to test a situational theory of crisis communi
cation. SCCT is premised on matching the crisis response to the
level of crisis responsibility attributed to a crisis. The study explores
one of the basic assumptions of SCCT by assessing whether the pre
dicted correlational relationship between crisis responsibility and
organizational reputation occurs across a range of crisis types.
Results support the theory’s predictions and suggest ways to refine
The organization’s reputation is widely recognized as a valued
resource (Winkleman, 1999). Crises, or unpredictable events that
can disrupt an organization’s operations, threaten to damage orga-
nizational reputations. An organization’s communicative response
to a crisis can serve to limit and even to repair the reputational dam-
age. Unfortunately, we have just begun to unpack how to use com-
municative responses to protect reputations.
Scholars frequently recommend highly accommodative strate-
gies that feature apologies for the crisis (e.g., Benoit, 1995;
Sellnow, Ulmer, & Snider, 1998). However, the universal applica
tion of highly accommodative strategies is problematic because of
the legal and financial liabilities they incur. Apologies require an
organization to publicly accept responsibility for a crisis, thereby
weakening its legal position in the event of a lawsuit (Fitzpatrick,
1995; Tyler, 1997).
Needed is a situational approach to selecting crisis response
strategies, that is, what an organization says and does after a crisis
to protect the organizational reputation. Benson (1988) was the first
to propose a situational approach. He challenged scholars to under
stand how the crisis type (or situation) influenced the selection of
crisis response strategies. To meet this challenge, scholars must
identify the range of possible crisis types and crisis responses and
explain how crisis types affect the selection of crisis responses.
166 MANAGEMENT COMMUNICATION QUARTERLY / NOVEMBER 2002
Some research has appeared that begins to expose how crisis type
influences the effectiveness of crisis responses (e.g., Benoit, 1995;
Coombs, 1999b; Hearit, 1996).
The purpose of this article is to explain a comprehensive, pre
scriptive, situational approach for responding to crises and protect
ing the organizational reputation: the situational crisis communica
tion theory (SCCT). SCCT is an extension of Coombs’s (1995)
previous research on matching crisis response strategies to the cri
sis situation, while also building on the work of other crisis man
agement scholars. This article summarizes the research that led to
the development of SCCT, articulates the central propositions of
SCCT, and reports an initial test of a key proposition.
SCCT CONSTRUCTS AND ASSUMPTIONS
SCCT assumes that an organization’s reputation, that is, how the
organization is perceived by its publics, is a valued resource that is
threatened by crises. A strategic communicative response can best
protect the reputational resource by assessing the crisis situation
and selecting a crisis response strategy that fits the crisis situation.
Of course, there are concerns other than reputation to address in a
crisis, particularly public safety. In fact, our research and that of
others has suggested that providing instructing information, that is,
what publics need to know and do to protect themselves from the
crisis, is necessary before addressing reputational concerns
(Coombs, 1999a; Coombs & Holladay, 2001; Sturges, 1994). How
ever, the central focus of SCCT is how to manage organizational
reputation during a crisis.
The crisis manager begins the selection of a crisis response strat
egy by identifying the crisis type, which we conceptualize as the
frame that publics use to interpret the event. The objective is to eval
uate the attributions of personal control, or the organization’s abil
ity to control the event, and crisis responsibility, or how much the
organization is to blame for the event. Perceptions of crisis respon
sibility have proven to increase as attributions of personal control
intensify (Coombs, 1998). In fact, personal control and crisis
responsibility may be so highly correlated as to merit treating them
Coombs, Holladay / INITIAL TESTS 167
as essentially isomorphic. At any rate, the level of crisis responsi
bility is a key indicator of the potential reputational damage a crisis
might inflict (Coombs & Schmidt, 2000).
The crisis manager should select a crisis response strategy that is
appropriate for the amount of potential reputational damage a crisis
may inflict. The stronger the potential reputational damage the
more the crisis response strategy must try to accommodate the vic-
tim or victims, that is, those adversely affected by the crisis. Publics
will expect an organization to do more for victims of the crisis when
the organization is held more accountable for the crisis (Coombs,
1995). The crisis response strategies should then mitigate
reputational damage by demonstrating that the organization cares
for the victims and knows the proper way to behave, thereby meet-
ing public expectations. In support of this claim, Coombs and
Holladay (1996) found crisis response strategies to mitigate
reputational damage in the case of organizational misdeeds. Fig
ure 1 illustrates the predicted relationships among key variables in
In assessing crisis types the crisis manager might consult a pleth
ora of crisis typologies that have been created (e.g., Fearn-Banks,
1996; Lerbinger, 1997; Pearson & Mitroff, 1993). These typologies
attempt to categorize or differentiate the kinds of crises that organi
zations experience. Unfortunately, the typologies were created sep
arately from typologies of crisis response strategies. This lack of
integration prevents crisis managers from using the lists of crisis
types to guide the choice of crisis response strategies. SCCT orga
nizes the crisis types so that they can be integrated with crisis
response strategies. In an attempt to integrate the multiple lists in
168 MANAGEMENT COMMUNICATION QUARTERLY / NOVEMBER 2002
Figure 1: Variables and Relationships in the Situational Crisis Communication
the literature, Coombs (1999b) developed a master list of nine crisis
types that we have refined and extended for this study. Table 1
defines the various crisis types used in this study.
For this study, Coombs’s (1999b) list was expanded to 13 to
reflect two important variations in crises. First, accidents and prod
uct recalls were each further differentiated to reflect the fact that
such crises may be caused either by technical breakdowns (e.g.,
equipment failures) or human breakdowns (e.g., errors). Several
studies suggested that people will view human breakdowns as eas
ier to prevent than technical breakdowns (Heath, 1994; Mitroff,
Harrington, & Gai, 1996; Pauchant & Mitroff, 1992). The second
variation that led to our expanded list is a refinement of organiza
tional misdeeds. This category covers a wide range of behaviors.
Thus, three variations of misdeeds were included: (a) those involv
ing injury, (b) those involving no injury, and (c) those involving a
legal or regulatory violation (managerial misconduct).
Identifying the crisis type enables an initial assessment of the
amount of crisis responsibility that publics will attribute to a crisis
situation. Adjustments are then made to this initial assessment by
considering two factors, severity and performance history. Severity
is the amount of damage generated by a crisis including financial,
human, and environmental damage. Performance history refers to
the past actions or conduct of an organization including its crisis
history (whether an organization has had previous crises) and rela-
tionship history (especially how well or poorly it has treated stake
holders). Severity and performance history have proven to modify
perceptions of crisis responsibility for some crisis types (Coombs,
1998; Coombs & Holladay, 1996, 2001). As severity increases or
performance history worsens publics will attribute greater crisis
responsibility to the organization. Thus, SCCT suggests that initial
assessments of crisis responsibility based on crisis type should be
adjusted upward or downward depending on severity and/or perfor
After assessing the level of crisis responsibility, crisis managers
then choose a crisis response strategy appropriate to the level of cri
sis responsibility. Scholars have developed many lists of crisis
response strategies; Coombs (1999b) synthesized these lists into
eight crisis response strategies: (a) an attack on the accuser, in which
the crisis manager confronts the group or person that claims a crisis
Coombs, Holladay / INITIAL TESTS 169
170 MANAGEMENT COMMUNICATION QUARTERLY / NOVEMBER 2002
13 Crisis Types Used in the Study
1. Rumor: circulation of false information designed to harm an organization. Snapple fac
ing a rumor that the ship on the label of its iced tea drink is a slave ship and that Snapple
gives money to the Ku Klux Klan was used as the case.
2. Natural disaster: a naturally occurring event (an act of God) that damages an organiza
tion. A bank in Homestead, Florida slowly reestablishing operations after Hurricane
Andrew was used as the case.
3. Malevolence/product tampering: damage by an external agent against an organization.
The Sudafed tampering in Washington that killed two people was used as the case.
4. Workplace violence: an attack by an employee or former employee on current employ
ees on the job. A shooting at the Radisson Bay Harbor Hotel Florida in which an
employee shot five coworkers was used as the case.
5. Challenge: confrontation by disgruntled stakeholders claiming an organization is oper
ating in an inappropriate manner. The American Family Association’s (AFA) picketing
of Waldenbooks was used as the case. The AFA maintained that Waldenbooks was a
pornography peddler because it carried Playboy and art books containing nudity. AFA
wanted all of these materials removed from the shelves of Waldenbooks.
6. Technical breakdown accident: an industrial accident caused by technology or equip
ment failure. The rupturing of an oil storage tank owned by Ashland Oil was used as the
case. The rupture was caused by flaws in the steel that went unnoticed during normal
inspections. The case notes that Ashland had no way of knowing about the flaws.
7. Technical breakdown product recall: the recall of a product because of technology or
equipment failure. A recall of 70,000 television sets by Sharp was used as the case. The
televisionsbecame fire hazards overtime as a faulty connector wore out. The faulty con-
nector, provided by a supplier, was not discovered until the overheating problem
emerged years later.
8. Megadamage: a technical breakdown accident that produces significant environmental
damage. A 210,000-gallon oil spill in Galveston, Texas by Buffalo-Marine Service, Inc.
was used as the case. One oil slick was over 5 miles long, and the spill posed a threat to
sea birds living in nearby wetlands. The environmental damage dominates the interpre
tation of the crisis, not the fact that it is a technical breakdown accident.
9. Human breakdown accident: an industrial accident caused by human error. The explo
sion of a chemical storage tank at a Ford truck plant was used as the case. Workers acci
dentally hooked up a line delivering phosphoric acid to a tank holding sodium nitrate.
The two solutions formed nitrous oxide and the pressure from the reaction blew the top
off the tank, releasing a gas cloud. Even if environmental damage occurs, the human
breakdown aspect of the accident becomes the defining feature of the crisis.
10. Human breakdown product recall: a product recall because of human error. A ham
burger recall by Hudson Foods was used as the case. The beef became contaminated by
E. coli when plant employees mistakenly placed contaminated beef back into the grind
ing process over a 3-day period. Their actions prompted the large product recall.
11. Organizational misdeeds with no injuries (to external stakeholders): management
knowingly deceives stakeholders but without causing injury. Chrysler disconnecting
odometers on test cars but not telling the buyers about the process was used as the case.
Customers did not know the real mileage on the test cars nor did they know the cars had
been driven hard test miles.
12. Organizational misdeed management misconduct: management knowingly violates
laws or regulations. Astra Pharmaceutical’s three top managers being found guilty of
gross sexual harassment of female sales representatives was used as the case.
exists; (b) denial, in which the crisis manager claims that there is no
crisis; (c) excuse, in which the crisis manager attempts to minimize
organizational responsibility for the crisis; (d) victimization, in
which the crisis manager reminds stakeholders that the organiza
tion is a victim of the crisis as well; (e) justification, in which the cri
sis manager attempts to minimize the perceived damage inflicted by
the crisis; (f) ingratiation, in which the crisis manager praises stake-
holders and reminds them of the past good works done by the orga-
nization; (g) corrective action, in which the crisis manager tries to
prevent a repeat of the crisis and/or repair the damage done by the
crisis; and (h) full apology, in which the crisis manager publicly
accepts responsibility for the crisis and requests forgiveness from
The eight crisis response strategies can be ordered along a con-
tinuum ranging from defensive, putting organizational interests
first, to accommodative, putting victim concerns first (Marcus &
Goodman, 1991; Siomkos & Shrivastava, 1993). The defensive-
accommodative continuum is adapted from the work of McLauglin,
Cody, and O’Hair (1983), who used the continuum to conceptual
ize accounts, or explanations people offer for their untoward or
Using this continuum the crisis manager then matches the crisis
response to level of crisis responsibility. The greater the crisis
responsibility generated by the crisis the more accommodative the
crisis response strategies must be. Following this principle should
offer maximum protection for the organizational reputation. The
very limited research thus far supports this recommendation
(Coombs & Holladay, 1996; Coombs & Schmidt, 2000). Crises
that prompt little to no attributions of organizational crisis respon
sibility, such as natural disasters, workplace violence, or rumors,
Coombs, Holladay / INITIAL TESTS 171
13. Organizational misdeeds with injuries (to external stakeholders): management know
ingly places stakeholders at risk and some are injured. Morton Feedmills letting dog
food be delivered that they knew contained small amounts of monensin, a poultry medi
cation, was used as the case. Monensin is toxic to dogs. An investigative report found
that managers allowed the shipments to occur, resulting in the nonfatal poisoning of a
number of dogs.
can be managed using just instructive information, telling people
what to do to protect themselves from the crisis, as the crisis
response. For rumors, crisis managers are recommended to add the
denial crisis response strategy, the most defensive strategy. Crises
that prompt moderate to low attributions of crisis responsibility,
such as accidents, are effectively managed via moderately defen
sive crisis response strategies such as excuse. Crises with strong
attributions of organizational crisis responsibility, such as organi
zational misdeeds, require strongly accommodative responses such
as corrective action and full apologies (see Coombs, 1995 or 1999b,
for a complete discussion of the matching process).
Extant research has tested some of the theory’s claims using a
limited range of crisis types. However, to refine the theory these
claims need to be tested across a broader spectrum of crisis types
and other claims need to be tested. The study reported here is a first
step in that direction.
HYPOTHESIS AND RESEARCH QUESTIONS
Previous research has treated personal control and crisis respon-
sibility as two separate but related variables; that is, researchers
have assumed that if an organization is perceived as being able to
control a crisis, then it will be perceived to have more responsibility
for that crisis (e.g., Coombs, 1998; Coombs & Holladay, 1996).
However, it is conceivable that personal control and crisis responsi
bility, although conceptually distinct, may be indistinct in their
operationalization or that measurements of the two variables may
tap the same underlying variable. Measures typically used to assess
personal control and crisis responsibility are derived from different
scales, but it may be appropriate to combine them into one compos
ite scale (Coombs & Holladay, 1996; Griffin, Babin, & Darden,
1992; McAuley, Duncan, & Russell, 1992). Thus, our first research
Research Question 1: Do the items for measuring personal control and
crisis responsibility measure two separate factors or one common
172 MANAGEMENT COMMUNICATION QUARTERLY / NOVEMBER 2002
As we assess crisis types it is relevant to consider how these cri
sis types might be grouped into clusters of like crises. A basic
assumption underlying much of crisis management planning is that
“crises can be grouped according to their underlying structural sim
ilarity” (Mitroff, 1988, p. 16; see also Marcus & Goodman, 1991;
Pearson & Mitroff, 1993). Organizations can use the crisis clusters
to construct crisis portfolios, that is, a crisis management plan for
each cluster to which the organization is vulnerable. The rationale
for clustering is that if an organization prepares a crisis plan for one
crisis in the cluster, it is reasonably well prepared for all crisis types
in that cluster (Mitroff, 1988; Mitroff et al., 1996). Crisis portfolios
are efficient because an organization may not have the time to
develop plans for every major crisis type and subvariation it may
Thus, it is worthwhile to explore how the crisis types in SCCT
might be grouped into crisis clusters. Because crisis responsibility
is central to assessing crisis types for the purpose of determining
strategic responses, it should be used to form the crisis clusters.
Thus, the second research question we pose is:
Research Question 2: What clusters will emerge from the 13 crisis
types when crisis responsibility is used as the cluster factor?
The relationship between crisis responsibility and organiza-
tional reputation is critical to SCCT. SCCT prescribes that crisis
managers utilize progressively more accommodative strategies as
crisis responsibility increases, because perception of crisis respon
sibility is believed to be directly correlated to reputational damage.
That is, as crisis responsibility increases, so does the reputational
damage inflicted by a crisis. Research has found support for the cri
sis responsibility-reputation relationship among a limited subset of
the crisis types (Coombs & Holladay, 1996; Coombs & Schmidt,
2000). It is instructive to determine if the relationship is found across
a wider array of crises. We would expect the relationship to exist in
all of the crisis clusters discovered through Research Question 2.
Hypothesis 1: Observers rate an organization’s reputation more nega
tively when they attribute greater amounts of crisis responsibility to
the organization, regardless of the crisis cluster.
Coombs, Holladay / INITIAL TESTS 173
The respondents in this study were 130 undergraduate students
enrolled in communication courses at a midwestern and a south
eastern university. Of the respondents, 64% were women (n = 83)
and 36% were men (n = 47). The respondents ranged in age from 18
to 54 years old (M = 21.8, SD = 5.40).
Three scales were employed in the study: Organizational Repu-
tation, Personal Control, and Crisis Responsibility.
Organizational Reputation. Organizational Reputation was
measured using five items from Coombs and Holladay’s (1996) 10-
item Organizational Reputation Scale, which is an adaptation of
McCroskey’s (1966) scale for measuring ethos. Specifically, an
adaptation of the Character subscale of McCroskey’s Ethos Scale
was used to assess organizational reputation. Character is concep-
tualized as the trustworthiness and good will of the source, that is,
an assessment of the degree to which the source is concerned with
the interests of others. Coombs and Holladay (1996) modified
McCroskey’s items by simply replacing the term speaker with
organization. The five items used in the present study were: (a)
“The organization is concerned with the well-being of its publics,”
(b) “The organization is basically DISHONEST,” (c) “I do NOT
trust the organization to tell the truth about the incident,” (d)
“Under most circumstances, I would be likely to believe what the
organization says,” and (e) “The organization is NOT concerned
with the well-being of its publics.”
In previous research the 10-item version of the scale had a
Cronbach’s alpha of .82 (Coombs & Holladay, 1996) and .92
(Coombs, 1998). To reduce the length of the survey, five items were
selected by reviewing previous research employing the Organiza
tional Reputation Scale and selecting those items that produced a
174 MANAGEMENT COMMUNICATION QUARTERLY / NOVEMBER 2002
Cronbach’s alpha higher than .80. The five-item scale used in the
present study had a Cronbach’s alpha of .87.
CEOs rate trustworthiness as one of the most important aspects
of an organizational reputation. Additionally, trustworthiness is a
common factor used in the commercial reputational measures,
including the reputational quotient (RQ), the most popular measure
for corporations (Fombrun, 1996; Winkleman, 1999). In fact, trust
worthiness is the primary construct tapped by the RQ. Hence, trust
worthiness is a viable measure for reputation even though it is
Personal Control. Personal control refers to the degree to which
an event is controllable or uncontrollable by the organization. It
was measured using four items inspired by the Causal Dimension
Scale II (CDSII) (McAuley et al., 1992). The items are: (a) “The
cause of the crisis was something the organization could control,”
(b) “The cause of the crisis is something over which the organiza-
tion had no power,” (c) “The cause of the crisis is something that
was manageable by the organization,” and (d) “The cause of the cri-
sis is something over which the organization had power.” Three pre-
vious studies have used this scale and reported Cronbach alphas of
.84 to .89 (Coombs, 1998, 1999a; Coombs & Schmidt, 2000). The
Personal Control scale in the present study had a Cronbach’s alpha
Crisis Responsibility. Crisis responsibility was measured using
Griffin, Babin, and Darden’s (1992) three-item scale for Blame.
The three items were: (a) “Circumstances, not the organization, are
responsible for the crisis,” (b) “The blame for the crisis lies with the
organization,” and (c) “The blame for the crisis lies in the circum
stances, not the organization.” Three crisis studies have used the
three-item Blame scale, with Cronbach alphas ranging from .80 to
.86 (Coombs, 1998, 1999a; Coombs & Holladay, 2001). The Crisis
Responsibility scale in the present study had a Cronbach’s alpha
The instrument also included four other items that were to be
used in a later study. The anchors for the Personal Control scale
were 1 (strongly disagree)to9(strongly agree). All other items had
anchors of 1 (strongly disagree)to5(strongly agree).
Coombs, Holladay / INITIAL TESTS 175
A total of 13 crisis cases were selected for the study, one for each
crisis type listed in Table 1. All crises were based on actual events
and used the real organizations’names as well as descriptions of the
crises provided by news reports.
Each respondent received a packet containing a cover page with
directions, two stimulus crisis cases, and two copies of the survey
instrument. The order of the materials in the packet was cover page,
first stimulus, first copy of survey instrument, second stimulus, and
second copy of the survey instrument. Each respondent read the
first stimulus and completed the first copy of the survey, then read
the second stimulus and completed the second copy of the survey
instrument. The two cases were randomly paired and then placed in
the packet. Respondents were verbally instructed to read carefully
each case and then respond to the questions following the case. The
administration required about 15 to 20 minutes.
Research Question 1, focusing on the possible isomorphic rela
tionships between personal control and crisis responsibility, was
explored using factor analysis. A principle components analysis
with varimax rotation was run using the items from the Personal
Control and Crisis Responsibility scales. Only one item had an
eigenvalue of over 1, so no rotation was needed. The results are pre
sented in Table 2. The single factor had an eigenvalue of 4.79 and
accounted for 67.98% of the variance. The results indicate that the
Personal Control and Crisis Responsibility scales seem to measure
the same factor. The two scales were combined to form a composite
score that will be referred to subsequently as Crisis Responsibility.
The Cronbach’s alpha for the new scale was .90. The new Crisis
176 MANAGEMENT COMMUNICATION QUARTERLY / NOVEMBER 2002
Responsibility scale was used in subsequent analyses in place of
the Personal Control and Crisis Responsibility composite scores.
Research Question 2, focusing on the groupings of crisis types,
was explored using hierarchical cluster analysis, an exploratory
method for discovering relatively homogeneous clusters of cases
based on some measured characteristics. It sorts cases, in this study
crisis types, into groups or clusters so that the degree of association
is stronger between members of the same cluster and weaker
between members of different clusters. A successful hierarchical
cluster analysis requires two important decisions: (a) identifying
variables used to create the clusters and (b) selecting a method for
determining the optimum number of clusters. Crisis Responsibility
was the variable used to create the crisis clusters because it is cen-
tral to SCCT. Crisis Responsibility and Organizational Reputation
scores were selected as the basis for evaluating the optimum num
ber of clusters. One-way ANOVAs were used to determine the opti
mum number of crisis clusters. The best fit occurred when the clus
ters were shown to be distinct from one another when compared on
the Crisis Responsibility and Organizational Reputation variables.
The cluster analysis found a four-cluster solution after two
stages and a three-cluster solution after three stages. The crisis
types in each cluster were combined to create cluster scores, then a
series of one-way ANOVAs was used to determine which cluster
solution created a best fit. Table 3 provides the results of the one-
way ANOVA analyses. The three-cluster solution seemed to best
represent the data in that the clusters were all significantly different
from one another on their scores for Crisis Responsibility and
Organizational Reputation; this was not the case in the four-cluster
Coombs, Holladay / INITIAL TESTS 177
Factor Analysis for Personal Control and Crisis Responsibility Items
Item Factor Loading
Personal Control 1 .89
Personal Control 2 .83
Personal Control 3 .75
Personal Control 4 .83
Crisis Responsibility 1 .86
Crisis Responsibility 2 .83
Crisis Responsibility 3 .85
TABLE 3: One-Way ANOVA Results for Crisis Clusters and Situation Crisis Communication Theory Variables
Natural Disaster Workplace Violence
Accidental Preventable Rumor Product Tampering
Variable Mean SD Mean SD Mean SD Mean SD F df p
Organizational Reputation 3.68
0.80 74.88 3,256 .001
Crisis Responsibility 4.24
1.27 30.54 3,256 .001
Accidental Preventable Victim
Mean SD Mean SD Mean SD
Crisis Responsibility 4.24
1.21 112.31 2,257 .001
Organizational Reputation 3.67
0.74 41.69 2,257 .001
NOTE: For each test, means superscripted a, b, and c are significantly different using Dunnett C procedure, p < .01.
solution. Hence, the 13 crisis types can be reduced to three crisis
clusters: the victim cluster, the accidental cluster, and the prevent
In all of the crisis types within the victim cluster the organization
is a victim of the crisis along with the stakeholders. This cluster
includes natural disasters, rumors, workplace violence, and prod
uct tampering. These are crisis types that produce minimal attribu
tions of crisis responsibility.
The second cluster was called accidental because all of the crises
represent unintentional actions by the organization; that is, the
organizations did not intend to create the crises. The accidental
cluster includes challenges, megadamage, technical breakdown–
accidents, and technical breakdown–recalls, that is, crisis types that
produce moderate attributions of crisis responsibility.
The third cluster was called preventable because the crises
involved either purposefully placing stakeholders at risk, or know-
ingly taking inappropriate actions, or human error that might have
or could have been avoided. The preventable cluster includes
human breakdown accidents, human breakdown recalls, organiza-
tional misdeeds–management misconduct, organizational misdeed
with no injuries, and organizational misdeeds with injuries, the cri-
sis types that produce strong attributions of crisis responsibility.
Hypothesis 1, focusing on the relationship between Crisis
Responsibility and Organizational Reputation, was tested using
Pearson correlation analysis. The Crisis Responsibility and Orga
nizational Reputation variables were correlated separately for each
of the three crisis clusters. The correlations were significant in all
three clusters: r = –.51 (p < .01) in the victim cluster, r = –.32 (p <
.01) in the accident cluster, and r = –.46 (p < .01) in the preventable
cluster. The results thus support Hypothesis 1.
An organization’s reputation is a valuable resource that should
be protected from the threats posed by a crisis (Barton, 2001).
SCCT offers a set of principles that guide the selection of crisis
response strategies in order to maximize reputational protection.
Coombs, Holladay / INITIAL TESTS 179
SCCT provides crisis managers with guidelines for understanding
which crisis response strategies are most appropriate for a given
crisis type (Coombs, 1995). Efforts to refine SCCT and to examine
a key assumption were tested in this study.
Using crisis responsibility to form crisis clusters, that is, group
ings of similar crisis types, refined thinking about crisis types.
Developing crisis clusters is premised on the logic of crisis portfo
lios: Similar crises can be managed in similar fashions (Pearson &
Mitroff, 1993). Because crisis types within each crisis cluster will
produce similar attributions of crisis responsibility, crisis managers
can use similar crisis response strategies to address crisis types
within the same cluster. Thirteen crisis types formed three distinct
clusters: the victim cluster, the accidental cluster, and the prevent
able cluster. The victim cluster involves crisis types in which harm
is inflicted on the organization as well as on stakeholders. The acci-
dental cluster involves unintentional actions by an organization. In
these crisis types the organization does not intend for the crisis to
occur; rather, the crisis situation results from a danger associated
with the organization’s operation. The preventable cluster, on the
other hand, involves intentionally placing stakeholders at risk,
knowingly violating laws or regulations, or not doing enough to pre-
vent an accident or a defective product from reaching the market.
Identifying the crisis type or placing an event in a crisis cluster is
the initial step in assessing crisis responsibility. The severity and
performance history factors must be considered as well. For exam
ple, repeated or severe technical breakdown accidents should per
haps be treated more like the preventable cluster than the accident
cluster. SCCT recommends that severity and performance history
factors be considered as part of the crisis responsibility adjustment
One additional point should be made about recalls and product
tampering crisis types, because they have a unique dynamic. Obvi
ously, recalls and product tampering will involve the corrective
action crisis response strategy, as the dangerous product must be
removed from the market. SCCT should be utilized to determine
what crisis response strategies to use in combination with the requi
site corrective action.
The model of SCCT presented in Figure 1 was made more parsi
monious by combining Personal Control and Crisis Responsibility.
180 MANAGEMENT COMMUNICATION QUARTERLY / NOVEMBER 2002
The factor analysis results indicate the two scales were measuring
the same factor. To retain the original language of SCCT, Personal
Control and Crisis Responsibility were collapsed into one variable
called Crisis Responsibility. Figure 2 is a revised model for SCCT.
The key assumption tested is that the relationship between crisis
responsibility and organizational reputation occurs across a range
of crisis types. The matching process of SCCT is premised on the
relationship between these two variables. Testing Hypothesis 1
found that a moderate correlation existed between crisis responsi-
bility and organizational reputation in all three crisis clusters. The
results suggest that this central relationship in SCCT can be applied
to all of the crisis clusters and is not limited to the organization mis-
deed crisis types examined in earlier research (Coombs &
Holladay, 1996, 2001; Coombs & Schmidt, 2000).
This study suffers from the limitations common to any study
using an experimental method and a student population. Although
the crisis events were real, respondents experienced the crises in an
artificial manner and not all the factors relevant to crisis manage
ment could be assessed. Furthermore, students are not the typical
targets for crisis communication. However, previous crisis commu
nication research has found no differences in responses between
student populations and nonstudent populations (Coombs, 1999a).
Moreover, students are and will continue to be stakeholders for
Coombs, Holladay / INITIAL TESTS 181
Figure 2: Revised Model of Situational Crisis Communication Theory
FUTURE RESEARCH DIRECTIONS
Research should begin to assess how people perceive the various
crisis response strategies. Researchers have arrayed crisis response
strategies from defensive to accommodative. Do stakeholders per
ceive crisis response strategies the way researchers believe they
will? Such research adopts a receiver perspective to understand
how people interpret crisis response strategies. Again, we must
make sure that our assumptions are accurate if we are using the
assumptions to recommend plans of action for crisis managers.
The results of Research Question 2 provide a solid base for mak
ing initial crisis responsibility assessments. The attributions of cri
sis responsibility for the crisis types and the crisis clusters provide
useful initial estimates that crisis managers can use to guide their
strategic selection of crisis response strategies. Future research
must unpack if the anticipated modifying effects of severity and
performance history are found across the spectrum of crisis types
and clusters. Currently, only technical breakdown accidents,
human error accidents, and organizational misdeed management
misconduct have been used to test the modifying effects of severity
and performance history (Coombs & Holladay, 1996, 2001, in
press). One question to ask would be, “Do severity and perfor-
mance history (crisis history and relationship history) have the
same effects on all crisis types in a crisis cluster?”
IMPLICATIONS FOR PRACTITIONERS
The SCCT is an attempt to provide crisis managers with a
resource for making informed decisions concerning ways to pro
tect the organizational reputation during a crisis. First, the identifi
cation of attributions of crisis responsibility for crisis types and the
formation of crisis clusters help crisis managers with their initial
assessment of crisis responsibility. Crisis managers can estimate
the level of crisis responsibility their crisis will generate among
publics by identifying the crisis type they face or by placing their
crisis in the appropriate crisis cluster. Once estimated, the level of
crisis responsibility serves to narrow the selection of viable crisis
response strategies. That is, crisis managers can make a more
182 MANAGEMENT COMMUNICATION QUARTERLY / NOVEMBER 2002
informed choice about which crisis response strategies to employ.
Combined with the results of this study, crisis managers may use
Coombs’s (1995) decision flowchart to help select the appropriate
crisis response strategy for each crisis cluster.
Second, the research found that the crisis responsibility–reputa
tion relationships occurred across a variety of crises. Because this
relationship is the foundation for the SCCT crisis response strategy
selection guidelines, crisis managers can place greater faith in the
Third, crisis managers have better insight into variations of acci
dent, recall, and organizational misdeed crisis types. For accidents
and recalls the distinction between technical and human break
downs is critical. Based on the results of the cluster analysis in this
study, it is apparent that stakeholders attribute much less crisis
responsibility to technical breakdown accidents and recalls than
human breakdown accidents and recalls. If the accident or recall
has its origins in a human breakdown, crisis managers must utilize
more accommodative strategies than for those with a technical
On the other hand, no differences were found in how people per-
ceive the three variations of organizational misdeeds in terms of cri-
sis responsibility or organizational reputation. This suggests that
any organizational misdeed must be treated seriously, and SCCT
recommends that highly accommodative crisis response strategies,
such as apology or corrective action, should be used.
Organizational reputations matter, and crises threaten organiza
tional reputations. Thus, crisis managers should make protecting
the organizational reputation a central focus of their work. SCCT
offers a way to explain and to practice organizational reputation
protection during a crisis. SCCT models the crisis process in terms
of crisis responsibility and organizational reputation and develops
a prescriptive system for matching crisis response strategies to the
crisis type. This study is the first in a series designed to test and to
refine SCCT. The study validated a key assumption in SCCT and
Coombs, Holladay / INITIAL TESTS 183
provided crisis managers with a basis for making more informed
choices of crisis response strategies in accidents, recalls, and orga
nizational misdeed crises.
Barton, L. (2001). Crisis in organizations II (2nd ed.). Cincinnati, OH: College
Benoit, W. L. (1995). Accounts, excuses, and apologies: A theory of image resto
ration. Albany: State University of New York Press.
Benson, J. A. (1988). Crisis revisited: An analysis of the strategies used by Tylenol
in the second tampering episode. Central States Speech Journal, 38, 49-66.
Coombs, W. T. (1995). Choosing the right words: The development of guidelines
for the selection of the “appropriate” crisis response strategies. Management
Communication Quarterly, 8, 447-476.
Coombs, W. T. (1998). An analytic framework for crisis situations: Better
responses from a better understanding of the situation. Journal of Public Rela-
tions Research, 10, 177-191.
Coombs, W. T. (1999a). Information and compassion in crisis responses: A test of
their effects. Journal of Public Relations Research, 11, 125-142.
Coombs, W. T. (1999b). Ongoing crisis communication: Planning, managing,
and responding. Thousand Oaks, CA: Sage.
Coombs, W. T., & Holladay, S. J. (1996). Communication and attributions in a cri-
sis: An experimental study of crisis communication. Journal of Public Rela
tions Research, 8, 279-295.
Coombs, W. T., & Holladay, S. J. (2001). An extended examination of the crisis
situation: A fusion of the relational management and symbolic approaches.
Journal of Public Relations Research, 13, 321-340.
Coombs, W. T., & Holladay, S. J. (in press). Matching crisis response strategies to
crisis situations: An attribution theory-based approach to crisis management.
In D. P. Millar & R. Heath (Eds.), Crisis communication: Theory and applica
tions. Mahwah, NJ: Lawrence Erlbaum.
Coombs, W. T., & Schmidt, L. (2000). An empirical analysis of image restoration:
Texaco’s racism crisis. Journal of Public Relations Research, 12, 163-178.
Fearn-Banks, K. (1996). Crisis communications: A casebook approach. Mahwah,
NJ: Lawrence Erlbaum.
Fitzpatrick, K. R. (1995). Ten guidelines for reducing legal risks in crisis manage
ment. Public Relations Quarterly, 40(2), 33-38.
Fombrun, C. J. (1996). Reputation: Realizing value from the corporate image.
Boston: Harvard Business School Press.
184 MANAGEMENT COMMUNICATION QUARTERLY / NOVEMBER 2002
Griffin, M., Babin, B. J., & Darden, W. R. (1992). Consumer assessments of
responsibility for product-related injuries: The impact of regulations, warn
ings, and promotional policies. Advances in Consumer Research, 19, 870-877.
Hearit, K. M. (1996, Fall). The use of counter-attack in apologetic public relations
crises: The case of General Motors vs. Dateline NBC. Public Relations Review,
Heath, R. L. (1994). Management of corporate communication: From interper
sonal contacts to external affairs. Hillsdale, NJ: Lawrence Erlbaum.
Lerbinger, O. (1997). The crisis manager: Facing risk and responsibility.
Mahwah, NJ: Lawrence Erlbaum.
Marcus, A. A., & Goodman, R. S. (1991). Victims and shareholders: The dilem
mas of presenting corporate policy during a crisis. Academy of Management
Journal, 34, 281-305.
McAuley, E., Duncan, T. E., & Russell, D. W. (1992). Measuring causal attribu
tions: The revised Causal Dimension Scale (CDII). Personality and Social
Psychology Bulletin, 18, 566-573.
McCroskey, J. C. (1966). An introduction to rhetorical communication.
Englewood Cliffs, NJ: Prentice Hall.
McLauglin, M. L., Cody, M. J., & O’Hair, H. D. (1983). The management of fail-
ure events: Some contextual determinants of accounting behavior. Human
Communication Research, 9, 208-224.
Mitroff, I. I. (1988, Winter). Crisis management: Cutting through the confusion.
Sloan Management Review, 29, 15-20.
Mitroff, I. I., Harrington, K., & Gai, E. (1996, September). Thinking about the
unthinkable. Across the Board, 33(8), 44-48.
Pauchant, T. C., & Mitroff, I. I. (1992). Transforming the crisis-prone organiza-
tion: Preventing individual, organizational, and environmental tragedies. San
Pearson, C. M., & Mitroff, I. I. (1993). From crisis prone to crisis prepared: A
framework for crisis management. The Executive, 7, 48-59.
Sellnow, T. L., Ulmer, R. R., & Snider, M. (1998). The compatibility of corrective
action in organizational crisis communication. Communication Quarterly, 46,
Siomkos, G., & Shrivastava, P. (1993). Responding to product liability crises.
Long Range Planning, 26(5), 72-79.
Sturges, D. L. (1994). Communicating through crisis: A strategy for organiza
tional survival. Management Communication Quarterly, 7, 297-316.
Tyler, L. (1997). Liability means never being able to say you’re sorry: Corporate
guilt, legal constraints, and defensiveness in corporate communication. Man
agement Communication Quarterly, 11, 51-73.
Winkleman, S. (1999, April). The right stuff. Chief Executive, 143, 80-81.
Coombs, Holladay / INITIAL TESTS 185
Timothy W. Coombs (Ph.D., Purdue University) teaches at Eastern Illinois
University and is a consultant for Communication Resources Northwest.
His work in crisis management and business continuity has appeared in a
variety of academic and professional journals. His main area of research is
crisis management/communication. He is the author of the award-winning
book Ongoing Crisis Communication.
Sherry J. Holladay is an associate professor at Eastern Illinois University,
Charleston. Her research interests include crisis communication, leader
ship, and aging employees. Her work has appeared in Management Com
munication Quarterly, Journal of Public Relations Research, and Interna
tional Journal of Aging and Human Development.
186 MANAGEMENT COMMUNICATION QUARTERLY / NOVEMBER 2002