Single-Motive and Multi-Motive Processing of a
Threat Appeal: Promoting the Preventative Health Behavior of
Running head: SINGLE- AND MULTI-MOTIVE PROCESSING
Jason W. Anderson
University of Wisconsin-Madison
Correspondence may be addressed to Jason Anderson,
Department of Communication Arts, University of Wisconsin-Madison, 821
University Avenue, Madison, Wisconsin, 53706, (608) 226-0431, or
This study considers the impact of behavioral commitment on the
cognitive and affective processing of a persuasive message advocating
influenza vaccination behaviors, and the resulting impact on the
integration of information into attitudes, behavioral intention, and
behavior. It was argued that prior behavior commitments would lead
some processors to engage in concurrent validity-seeking and defensive
processing. This multi-motive processing may explain the limited
effectiveness of persuasive messages, specifically threat appeals. A
non-random sample (n=178) of university students was collected.
Results with regard to attitude ambivalence (p<.05), and other
indicators (i.e., negatively valenced cognitions, biased cognitive
processing, and affective arousal) were consistent with predicted
differences between single- and multi-motive processors. Implications
for theoretical and applied research are discussed.
multi-motive processing, defensive processing, validity processing,
persuasion, and influenza.
Single-Motive and Multi-Motive Processing of a
Threat Appeal: Promoting the Preventative Health Behavior of
Many attempts at persuading individuals to engage in
healthy behaviors are based on the assumption that making one aware of
a life-threatening event should lead to the adoption of behaviors that
alleviate the threat. These persuasive messages, known as threat
appeals (Leventhal, 1971; Rogers, 1975), have been shown to effective
in garnering behavioral change in a variety of contexts, including
health messages (Boster & Mongeau, 1984; Mongeau, 1998). In his
meta-analysis, Mongeau (1998) reported correlations of .19 between
level of threat and attitude change and .12 between level of threat
and behavioral change. These analyses support threat appeals as a
compelling message form and means of garnering attitude and behavior
change. As such, threat appeals are an important tool at the disposal
of health message designers. On the other hand, the correlations
observed in the meta-analyses are still far from universally effective
even though goals for self-preservation are surely the most
fundamental of all human motivations.
In order to clarify the limited efficacy of threat
appeals, this study examines individuals exposed to a persuasive
message—a threat appeal—concerning influenza and the need for
influenza vaccinations. The findings detail the effects of motivations
for self-preservation, and in some instances, the effects of
countervailing motivations that serve to constrain the effectiveness
of threat appeals. Moreover, the study provides an insight into when
these competing motives are present and how they affect the persuasion
Primarily spread through the air, influenza, commonly
known as "the flu," poses a threat to everyone in the proximity of the
carrier. On average, about 10% of all Americans contract the virus
each flu season (http://www.cdc.gov). The only cure for the virus is
to let it run its course, a process that may take two weeks (http://www.nfid.org).
As a result, most of these victims only incur the miseries that
accompany the infections (i.e., fever, muscle aches, fatigue, chills,
sweating, etc.), as well as possible work, school, and leisure-related
debts. These hardships are certainly unwanted, but many who contract
the disease lose much more. During an average year, approximately
20,000 Americans die from influenza and many more are hospitalized (http://www.cdc.gov;
American Medical Association, 1999). When influenza leads to influenza
related pneumonia, the flu becomes deadly. In fact, influenza and
influenza-related pneumonia are the sixth most common cause of death
in the United States (http://www.cdc.gov). Furthermore, influenza can
exacerbate existing medical conditions (i.e., asthma, heart disease,
emphysema, AIDS, diabetes) and lead to additional medical
University of Wisconsin Hospitals and Clinics, 1999). Some strains of
influenza have more dire consequences than others. The Spanish Flu
claimed the lives of 500,000 Americans and 20 million people worldwide
from 1918-1919 (http://www.cdc.gov). Researchers note that it is a
question of "when," not "if" another pandemic occurs, and that the
world is far overdue for another flu pandemic (http://www.onhealth.com).
While no cure for the flu exists, vaccinations, which decrease the
chances of contracting the flu from 20% to 2% and shorten recovery
times, are a viable tool for combating the flu (http://www.cdc.gov,
http://www.nfid.org, World Health Organization, 1999). Despite this
information, not all Americans receive annual vaccinations. Therefore,
this study seeks to understand and garner flu vaccination behaviors.
Threat Appeals as Persuasive Messages and Motivations To Avoid
Extant research concerning influenza vaccination behaviors
appears to focus on identifying predictors of vaccination behaviors
(Chapman & Coups, 1999) and the role of information-based
interventions in increasing vaccination rates (Herman, Speroff, &
Cebul, 1994; Ohmit, Furumoto, Monto, & Fasano, 1995) particularly
among at risk groups (i.e., the elderly). The research in this area
does not appear to take advantage of a body of literature concerning
the design of persuasive messages, including threat appeals. However,
existing research may still inform the design of persuasive messages
in this context.
Herman et al.’s (1994) information-based intervention found that
interventions that illustrated the dangers of failing to get a flu
shot for the elderly were somewhat successful in increasing rates of
vaccination, but the intervention was only successful in achieving a
36% vaccination rate compared to the 23% reported by the control
group. The findings illustrate the need for persuasive messages to go
beyond not only identifying threats to one’s health, but also
presenting a compelling argument for addressing the health threat.
Threat appeals are one form of persuasive messages that have been
shown to be effective (Mongeau, 1998), and in this study, a threat
appeal serves as a vehicle for conveying a persuasive message. By
their nature, threat appeals have implications for one’s
self-preservation goals, but threat appeal exposure does not lead to
universally practiced health behaviors (Mongeau, 1998). Therein lies
the focus of the present study.
Influenza infections can be deadly (http://www.cdc.gov),
and infections can have costly social and economic effects; however,
despite the undeniably high costs and likelihood of infection,
individuals still fail to perform health behaviors that can help one
avoid these costs (i.e., receive an influenza vaccination). Despite
salient threats to one’s goals for self-preservation, across various
health issues the research suggests that threat appeals are not as
effective in changing behaviors as it seems they should be (Mongeau,
1998). Within the context of influenza vaccination behaviors, this
study searches for explanations to these findings by identifying other
salient motivations that might conflict with goals for
Functional theorists hold that attitudes may serve
multiple functions. Smith, Bruner, and White (1956), as well as Katz
(1960) agree that one such function is to allow an individual to
contemplate an attitude object (e.g., get a flu shot) in order to
discern the potential harms or benefits the attitude object holds for
the individual. Herek’s (1986) neofunctional approach builds from this
contention. Herek contends that message targets process messages in an
effort to accurately evaluate the evidence and claims forwarded by a
persuasive message. This concern for accuracy is also reflected in the
development of the Heuristic Systematic Model (HSM) of message
processing (Chaiken, Eagly, & Liberman, 1989). According to the HSM,
message targets may process messages through a validity seeking
orientation in an effort to maintain, reformulate, or develop
“accurate” attitudes (Chaiken et al., 1989). Therefore, one motive for
message processing is the quest to hold accurate attitudes.
As suggested by functional theorists and HSM research,
individuals may hold other motives for evaluating a persuasive
message. Thus, for some individuals, a persuasive message (i.e., a
threat appeal) may inherently attack a held attitude or behavior
position. In such cases, one salient motive may be a defense motive.
Under such a motive, individuals engage in defensive processing in
order to protect an attitude or behavior in which they have a vested
interest or prior commitment (Chaiken et al., 1989; Giner-Sorolla, &
Chaiken, 1997; Herek, 1986). When individuals have already formed a
position on an issue, persuasive messages that attack the attitude or
behavior may motivate them to engage in defensive processing in an
effort to avoid any requisite changes in one’s self-image made
necessary by a successful attack.
Validity seeking and defense motives should result in
substantively different processing outcomes. Given a compelling
persuasive message, individuals processing through a validity seeking
orientation should show (a) a large number of cognitive responses, (b)
a low number of counterarguments, (c) strong agreement with the
advocated position, (d) a strong attitude-intention correlation, (e) a
strong intention-behavior correlation, and (f) low attitude
ambivalence. Relative to validity seeking processors, defensive
processors should show (a) roughly the same number of cognitive
responses, (b) a high (versus lower) number of counterarguments, (c)
weaker (versus stronger) agreement with the advocated position, (d) an
equally strong attitude-intention correlation, (e) an equally strong
intention-behavior correlation, and (f) roughly the same low level of
attitude ambivalence. In order to make sense of these motivations, it
is important to consider the potential conflict between them.
Single- Versus Multi-Motive
It has been argued that individuals exposed to a
persuasive message may orient toward both validity seeking and defense
motives with each motive yielding a unique impact on message
processing. However, it is also important to note the tension between
these potentially salient motives and their processing demands. For
influenza vaccinations and other health behaviors, the rewards of such
processing behaviors may conflict with potential rewards for
performing an alternative behavior. For example, condom use affects
one’s likelihood of contracting AIDS or an STD, yet a large portion of
the non-monogamous population fails to use condoms for a variety of
reasons (i.e., loss of sensation, lack of spontaneity in the sexual
act, etc.). Similarly, flu vaccinations may allow one to greatly avoid
the effects of an influenza outbreak. At the same time, getting an
influenza vaccination also has costs such as pain and resource debts
(e.g., loss of leisure time, money for the vaccination, etc.).
Furthermore, for those who have never received a vaccination, they may
be forced to re-evaluate their self-concept (e.g., I was wrong about
the need for vaccinations, I am not invincible, etc.).
Clearly there are reasons for persuasive health messages
to orient individuals to engage in validity seeking in an effort to
meet goals for self-preservation; however, for some individuals, there
appear to be reasons to orient toward a defense motive as well.
Cognitive consistency theories provide a rationale for individuals
orienting toward defensive processing in the face of a persuasive
health message (Festinger, 1957). These theories argue that
individuals desire to hold attitudes and perform behaviors that are
consistent with beliefs and behaviors performed in the past. Thus,
defensive processing allows one to avoid the tension or dissonance
that arises from performing behaviors or espousing attitudes that run
counter to prior behaviors and beliefs.
In traditional laboratory research, investigators tend to manipulate
conditions or assume that participants engage persuasive messages
under a validity seeking orientation. But, as cognitive consistency
theories suggest, the notion of single-motive validity seeking
processing seems unwarranted. In applied contexts, individuals exposed
to a message bring with them a history of issue-relevant behaviors and
attitudes. Therefore, the effectiveness of persuasive messages may be
affected by motivations that run counter to the message advocacy.
Given a compelling, persuasive message
advocating influenza vaccination behavior, one should expect all
message recipients to recognize threats to self-preservation goals and
orient toward validity seeking processing. And, for those who have a
history of receiving flu vaccinations and positive attitudes toward
vaccinations, the assumption that one engages in validity seeking,
single-motive processing appears warranted. But, not all individuals
have a history of practicing flu vaccination behaviors. In the context
of a persuasive message advocating influenza vaccinations, messages
might implicate not only a self-preservation goal, but also a held
attitude and behavior toward vaccinations. As a result, those with a
vested interest or behavioral commitment to vaccination avoidance may
realize conflict as they attempt to orientate toward defense and
validity seeking processing demands. If we can also assume that all
individuals seek self-preservation, then some message targets must
engage in multi-motive processing (i.e., concurrent defense and
validity seeking motivations) based on one’s prior attitudes and
beliefs toward the issue at hand. It is argued that, unlike
single-motive processors, multi-motive processors realize conflict
while engaging in message processing. The following discussion focuses
on how this conflict manifests itself in the persuasion process.
Manifestations of Single- and
Cognitive response. When individuals are motivated to process
messages through a desire to hold valid attitudes, there is an
increase in the depth of message processing (e.g., Chaiken et al.,
1989). Furthermore, defense motives have been shown to enhance one’s
depth of processing (Giner-Sorolla & Chaiken, 1997). Thus,
multi-motive processors might be expected to exhibit greater depth of
processing than single-motive processors given the joint effects of
the motives (i.e., concurrent validity seeking and defense motive
With regard to the issue of influenza vaccinations, when exposed to a
message favoring vaccinations, we suggest that:
H1: Non-vaccinators (i.e., behavioral commitment to avoid
vaccinations) will report more total cognitions than prior vaccinators
(i.e., behavioral commitment to get vaccinations).
Beyond the depth of processing, research suggests that
single- and multi-motive processors may differ in the valence of their
reported cognitions. Research suggests that cognitive reports that are
favorable toward the message (i.e., supporting arguments) result in
agreement with a message’s advocacy (Giner-Sorolla & Chaiken, 1997).
On the other hand, cognitive reports that are unfavorable (i.e.,
counter-arguments) result from a disagreement with the message’s
advocacy. Given their prior behaviors toward vaccinations and exposure
to a compelling persuasive message, we suggest that:
H2: Non-vaccinators will produce a more negative dominant
cognitive response (i.e., total positive cognitions minus total
negative cognitions) than prior vaccinators.
Biases in message processing. This study employs a
threat appeal as a form of a persuasive message. Threat appeals are
composed of two key elements (Dillard, 1994; Mongeau, 1998). First,
there is a threat component. The threat component orients the
message recipient to the negative outcomes associated with failing to
adopt the recommendations of the message. The threat component
contains two central elements: (a) severity and (b) likelihood
(Dillard, 1994; Witte, 1993, 1994). The severity element emphasizes
the nature of the threatening event and the costs of one’s present
behavior, while the likelihood element expresses the probability that
the event will occur and the consequences of the event for the
Second, the action component details the
recommendations that must be carried out to avert the threatening
event. Again, the action component contains two key elements: (a)
response efficacy and (b) self-efficacy. Response efficacy addresses
the likelihood that carrying-out the recommendations advocated by the
message will avert the threatening event, and self-efficacy emphasizes
one’s ability to carry-out the recommendations (Dillard, 1994; Witte,
The nature of each of these key elements of the threat appeal is
determined by a subjective judgment made by each message recipient. It
is proposed behavioral commitments lead individuals to engage in
defensive processing. As such, one’s perceptions of the severity,
likelihood, response and self efficacy may be biased by behavioral
commitments. Therefore, we propose that:
H3: Compared to prior vaccinators, non-vaccinators will
underestimate one or more of the following: (a) the perceived
likelihood of contracting influenza, (b) the perceived severity of
contracting influenza, (c) the response-efficacy of recommended action
(i.e., obtain an influenza vaccination), and (d) their self-efficacy
regarding the execution of the recommended action.
Emotional responses to the persuasive message.
While cognitive responses to persuasive messages are of interest, it
is also important to consider one’s emotional responses. Emotions
result from one’s appraisal of the interaction between one’s goals and
environment (Frijda, 1986; Lazarus, 1991). Specifically, when one
appraises a threat to self-preservation goals based on salient
features of his or her environment, fear is aroused (Lazarus, 1991).
Threat appeals are specifically designed to emphasize such threats
regardless of one’s behavioral commitments. Therefore, we propose that
validity seeking processors—single- and multi-motive processors
alike—will experience an arousal of fear. Specifically, we predict:
H4: Following exposure to the threat component of the
message, both prior vaccinators and non-vaccinators will experience an
increase in fear.
Similar to the appraisal process for fear, anger is aroused when
individuals recognize a presence in their environment that threatens
one’s goal for autonomy (Lazarus, 1991; Dillard, Kinney, & Cruz,
1996). For those individuals who have a vested interest or behavioral
commitment to avoid flu vaccinations, a threat appeal that attacks
these attitude and behavioral positions should be viewed as an attempt
to limit one’s autonomy (Brehm & Brehm, 1981; Lazarus, 1991; Dillard,
Kinney, & Cruz, 1996). Due to this nature of threat appeals, the
persuasive message should lead these individuals to not only
experience fear following measure exposure, but also anger. Therefore,
H5: Non-vaccinators will report a greater experienced
level of anger than prior vaccinators.
Researchers have also noted that threat appeals arouse emotions other
than fear and anger (Dillard, Plotnick, Godbold, Friemuth, & Edgar,
1996). Specifically, we were interested in differences in the arousal
of negative emotions among single- and multi-motive processors.
Differences in the arousal of sadness and guilt might explain
differences obtained between these groups of processors. Therefore, we
RQ1: What is the relationship between the
experience of sadness and guilt following exposure to the threat
component of the message for prior vaccinators and non-vaccinators.
While it is important to consider the
effects of message processing, it is also important to consider the
resultant impact of multi-motive processing on the integration of this
information into attitude, behavioral intention, and behavioral
outcomes. Below, these effects are addressed.
Information integration. According to combinatorial
theories of attitudes, attitude change occurs by integrating
information with pre-existing attitudes (Anderson, 1971; Fishbein &
Azjen, 1975). When exposed to messages containing information salient
to one’s attitudes, individuals weight the salient information and
then arrive at an attitude polarity judgment represented somewhere on
a continuum ranging from good to bad at the extremes (Fishbein & Azjen,
1975). Among single-motive processors, we expect that integration is
smooth and efficient. However, among multi-motive processors,
individuals are oriented toward two competing and conflicting motives,
validity seeking and defense. This conflict arises out of the desire
to defend an attitude through defensive processing, while concurrently
addressing one’s desire for self-preservation through rational,
unbiased validity processing. Therefore, we propose:
H6: Non-vaccinators will produce a correlation between
dominant cognitive response and attitude that is lower in magnitude
than prior vaccinators.
Attitude ambivalence. The conflict between these
two competing motives should also be realized in terms of the
certainty with which attitudes are held. While single-motive
processors are free to focus on evaluating the validity of the
message’s advocacy, multi-motive processors not only seek to evaluate
the validity of the message, but also to defend an attitude position
to which they are behaviorally committed (i.e., avoid influenza
vaccinations). Due to the concurrent motivations, multi-motive
processors exhibit less certainty in their attitude position, or in
other words, greater attitude ambivalence (Gross, Holtz, & Miller,
1995). Thus, we propose:
H7: Non-vaccinators will report higher levels of attitude
ambivalence than prior vaccinators.
Bridging the gap from attitudes to behaviors.
Fishbein and Azjen’s (1975) theory of reasoned action, as well as
Azjen’s (1991) theory of planned behavior, illustrate the role of
attitude polarity judgments as a cause of behavioral intention. The
linkages between attitude judgments and behavioral intentions have
also been established in meta-analyses of this relation (Kim & Hunter,
1993a, 1993b). Unlike single-motive processors, multi-motive
processors must make sense of information gleaned from conflicted
message processing. We propose that the conflict impacts the
attitude-behavioral intention relationship such that:
H8: Non-vaccinators will produce a correlation between
attitude and behavioral intention that is lower in magnitude than
Similarly, Fishbein and Azjen’s (1975) theory of reasoned action has
shown that behavioral intention is the best predictor of behavior—a
correlation of about .6. Again, for multi-motive processors, the
conflict that was generated at the information integration level of
processing has implications for not only the attitude-behavioral
intention relation, but also carry-over effects on the behavioral
intention-behavior relation. Specifically, we propose that:
H9: Non-vaccinators will produce a correlation between behavioral
intention and behavior that is lower in magnitude than prior
These hypotheses allow us to examine the persuasion process from
message exposure through the enactment of subsequent behaviors. This
examination provides an insight into both affective and cognitive
processes that take place among single- and multi-motive processors.
The data for this study were obtained as part of a larger
study conducted at a large Mid-western university.[ii]
The initial sample for this study consisted of 181 participants. In
order to remove any participants who were physically unable to receive
a flu shot, participants were asked to report whether or not they were
pregnant or allergic to eggs. This resulted in the removal of two
participants from the analyses. An additional participant had to be
excluded because she failed to report her vaccination history.
Therefore, the sample size for the analyses below is 178 including 53
males and 125 females with a mean age of 20.21 (range=17-34) (See
Table 1). Participants were recruited from undergraduate courses
across various departments at the university, and they received a
small amount of extra credit for completing the study.
Prior to beginning the experiment, two threat appeals were developed
with the intention of creating differential levels of threat. Given
the data subset of interest, the focus here is on the development of
the high threat message during the pre-test phase. We sought to
construct a message that was persuasive, compelling, and as realistic
as possible. To that end, the message was constructed to include each
of the necessary elements of a threat appeal (Dillard, 1994; Witte,
1993). Beyond constructing a threat and action component, we also
sought to develop compelling arguments within each component. These
arguments were supported by evidence from expert sources (e.g., Center
for Disease Control, National Institute of Infectious Diseases, etc.)
and were presented in both a narrative and a statistical form in an
effort to produce a compelling message (See Table 2). Morley (1987)
identified three indicators of compelling messages: (a) novelty, (b)
believability, and (c) importance. The messages were pre-tested along
each of these dimensions to determine whether or not the message was
compelling. Three close-ended items on 7-point semantic differential
scales were utilized to assess each argument. Pre-tests suggested that
the respondents (n=27) perceived the message to be compelling.
On average, the respondents found the arguments contained in the
message to be novel (X=4.35, range=1.53-6.30, sd=1.17),
believable (X=5.8, range=4.25-7.00, sd=.85), and
important (X=5.44, range=2.1-7.00, sd=6.1). With mean
scores consistently above the mid-point of scales, the study proceeds
under the assumption that the message can be characterized as
persuasive and compelling.
Procedures and Measures
The onset of data collection for this study coincided with that of a
University Health Service’s (UHS) program that provided free influenza
vaccinations to all interested university students, faculty, and staff
members. Data collection for this phase of the study continued until
one week before the end of the program.
Upon arrival at the laboratory, participants were randomly assigned to
a high or low threat messages condition (with the high threat
condition being the focus of the analyses below), as the experimenter
distributed a questionnaire, message booklet, and consent form to each
participant. Next, the experimenter read aloud the cover story
explaining the experimenters were interested in refining health
messages to be used in an actual web-based health campaign. Then,
participants were directed to sign a consent form if they still wished
to participate before continuing with the study. This phase of the
study took about half an hour to complete.
The second phase of the study consisted of two parts and began
immediately after the conclusion of UHS’ flu vaccination program.
During phase one of the study, we had asked participants to give us
consent to review their flu vaccination records on file at UHS. Their
consent allowed us to discern whether or not they had participated in
the UHS program. The other aspect of the second phase of the program
concerned a follow up questionnaire that was administered online via a
university website. The questionnaire contained self-report behavioral
measures that are discussed below. Collection procedures of phase two
are discussed below.
Measurement. In most persuasion research,
participants are presented with a complete message and asked to
provide their evaluations of it. However, threat appeals are generally
constructed such that there are two clear components to the message,
that is, the threat or problem component followed by the action or
solution component. We had an interest in evaluating each of these
components, while connecting our research to previous work. Thus, two
measurement conditions were created. First, the interrupted
measurement condition was created by asking participants to read the
threat component. After reading the threat, the participants then
provided closed- and open-ended data regarding their reactions to it.
At this point in the interrupted condition, respondents were asked to
return to the message booklet, read the action component and provide
data relevant to that section of the message. In the second condition,
labeled non-interrupted, participants read through the entire message
then provided data on each section individually. The experimental
instructions for the two conditions differed. Therefore, each
session was randomly designated as either an interrupted or a
non-interrupted. Commitment to Flu Vaccination Behavior.
In order to discern participants’ prior influenza vaccination
behaviors, a single-item measure was included in the questionnaire.
Specifically, participants were asked “How many times have you
received an influenza vaccination?” Response options included were as
follows: (a) never, (b) once or twice, (c) three-four times, and (d)
five or more. This item was then used to create a dichotomous flu
vaccination behavior variable that serves as the independent variable
in the analyses that follow. Participants reporting no history of flu
vaccinations (n=88) were labeled non-vaccinators, while
those reporting at least one prior flu vaccination (n=90) were
Open-Ended Responses. Following exposure to the
message, respondents were asked to list any thoughts and feelings they
had while reading the message. These reports were then evaluated by
three trained coders.[iii]
In the first pass through the data, the coders unitized the
responses. Participants were instructed to report one thought or
feeling per response box, but it was common to find reports such as "I
want to get [a flu shot] right away, because I am afraid I will
forget." These reports were best considered as two separate thoughts.
Therefore, 20 questionnaires were randomly selected. Working
independently, each coder unitized the data into psychological thought
units (roughly, independent clauses). Then, the percentage of
agreement for the first pass through the data was computed as twice
the number of agreements divided by Coder 1's total units plus Coder
2's total units for each pair of coders yielding agreement of 86%,
89%, and 84%. At this point, sources of error were discussed and
additional training was provided. When the above procedures were
repeated, a second pass through the data yielded acceptable agreement
levels of 98%, 97%, and 97% between the pairs of coders. Given these
levels of agreement, each coder proceeded to unitize a third of the
remaining questionnaires yielding a total of 5,515 open-ended
A second round of coding was conducted to identify
open-ended reports of affective experiences that were redundant with
close-ended reports of the same. Utilizing a list of feeling terms
compiled by Shaver, Schwartz, Kirson, and O'Connor (1987), coders
classified a unit as affective whenever these words appeared within a
participant's report. Coding of a random sample of 20 questionnaires
yielded acceptable reliability (κ= .96, .93, and .94), thus further
training was not warranted. After coding, the affective units were
removed, and a data set of 5,099 cognitive responses was obtained.[iv]
Finally, coders classified the remaining responses as
either (a) supportive comments (i.e., responses expressing
agreement with the message), (b) negative comments (i.e.,
responses expressing disagreement with the message), and (c)
neutral comments (i.e., response that merely reiterates the
message). Although initial coding attempts were not acceptably
reliable (κ=.76, .78, and .84), reliability was established after
additional training that addressed problem areas (κ=.87, .85, and
.95). At this point, the remaining data set was coded with 1,405
supportive, 607 negative, and 3,087 neutral comments identified.
The open-ended data were utilized to create two measures:
total cognitive response and dominant cognitive response.
Total cognitive response was created by summing all negative,
supportive and neutral comments produced by each participant. Dominant
cognitive response was created by subtracting all the negative reports
produced by a participant from all the positive reports produced.
Essentially, dominant cognitive response measures that approach or are
less than zero indicate open-ended responses that are increasingly
counter-argumentative in nature.
Severity. For the severity variable, three
close-ended items on a 5-point Likert-type scale ranging from strongly
disagree (1) to strongly agree (5) were included (e.g., “From what I
know about the flu, I think that it is very serious”).
Alpha for the three-item measure was .63.[v]
Attempts to improve the reliability were unsuccessful.
Probability. We also assessed the perceived
probability of contracting influenza. Three close-ended items on a
5-point Likert-type scale ranging from strongly disagree (1) to
strongly agree (5) were included (e.g., "There is a real possibility
that I could contract influenza this academic year"). Cronbach's alpha
for the measure was .84.
Response efficacy. Reponse efficacy was collected
using three close-ended items with a 5-point Likert-type scale ranging
from strongly disagree (1) to strongly agree (5) (i.e., "Getting a
vaccination is a sure-fire way to reduce the possibility of
contracting the flu"). Reliability analyses revealed a low reliability
level (α=.52) with the three-item measure, but when one of the items
was dropped, a reliability level of .72 was achieved and used in
Self-Efficacy. To assess self-efficacy, we
utilized three close-ended items with a 5-point scale ranging from
strongly disagree (1) to strongly agree (5) (i.e., “If I resolved to
get a flu shot, I am certain that I would be able to do it"). Again,
preliminary analyses yielded an unacceptable reliability level (α=.50)
for the three-item measure, but when one item was dropped the alpha
improved to an acceptable .81. This two-item measure was used for the
Emotions. Emotion measures of (a) fear, (b) anger,
(c) sadness, (d) guilt, and (e) happiness were collected using items
that had been demonstrated to be reliable (Dillard & Peck, 1998). The
scales for each of the emotion items ranged from none of the emotion
experienced (0) to a great deal of the emotion experienced (4). Three
items were collected for measures of fear (i.e., fearful, afraid, and
scared), anger (i.e., irritated, angry, and aggravated), and sadness
(i.e., sad, dreary, and blue), while the guilt measure consisted of
two items (i.e., guilty and ashamed). Respondents were asked to
respond to the emotion items before reading the threat appeal (i.e.,
baseline) and after reading the threat appeal (i.e., post-threat).
Analyses revealed that baseline emotion measures for fear (α=.82),
anger (.84), sadness (.67), and guilt (.72), as well as post-threat
measures (α=.94, .90, .74, and .70 respectively) were reliable. Using
these measures, we developed an emotion change score for fear, anger,
sadness, and guilt by subtracting the baseline measure of each emotion
from the post-threat measure.
Attitude. Participants were exposed to a message
that urged them to take part in the UHS’ free influenza vaccination
program. After reading the entire message, they were asked “The idea
of me getting a flu vaccination at University Health Services is…”
followed by a series of seven 7-point semantic differential scales
(i.e., good/bad, wise/foolish, positive/negative,
favorable/unfavorable, undesirable/desirable, necessary/unnecessary,
and not essential/essential). The attitude toward the message advocacy
measure was reliable (α=.92).
Behavioral intention. Following exposure to the
entire message, respondents were asked to make a probability judgment
concerning their likelihood of getting a flu vaccination. We asked
“All things considered, how likely is it that you will get a flu
vaccination from University Health Services during the 1999-2000
school year?" Participants reported their judgment on a scale ranging
for (0) “Certain that I will not” to (100) “Certain that I will.”
Attitude ambivalence. This variable was
constructed from three close-ended items using five-point Likert-type
scales ranging from (1) strongly disagree to (5) strongly agree (e.g.,
“When thinking about the flu, my mind is split on whether I should or
should not get a vaccination"). Cronbach’s alpha for attitude
ambivalence was .61. Attempts to enhance the reliability were
unsuccessful. Therefore, lack of support for findings regarding the
attitude ambivalence variable should be interpreted with caution.
Behavior. A behavior measure was created from two
sources of data. First, during phase one of the study, some
participants gave the researchers informed consent to review UHS flu
vaccination records. Therefore, records were reviewed and behavioral
measures were coded as either (a) having received a vaccination
following exposure to the threat appeal, (b) having received a
vaccination prior to exposure to the threat appeal, (c) not receiving
a vaccination, or (d) not consenting to review of records. This coding
resulted in a total of 33 missing cases on the behavior variable.
Subsequently, the self-report data from the questionnaire administered
during the second phase of the study were reviewed for these 33
missing cases. By cross-referencing the self-report data with the
other behavioral data, six missing data points were eliminated from
the behavior variable. This procedure yielded a total sample size of
n=152 for the behavior variable.
Descriptive Analyses. Before any of the main analyses were
interpreted, the distributions of the dependent variables were
analyzed, but the findings did not signal any immediate problems with
the variables of interest.
Measurement Condition. In order to enable the main
analyses below to ignore the measurement manipulation, a multivariate
analysis was conducted to test for any main or interaction effects of
the measurement condition upon our dependent variables of interest.
Neither a main effect for measurement condition (Λ =.959, ns),
nor an interaction effect with our independent variable—vaccination
history—(Λ=.921, ns) was observed. Therefore, the effects of
the measurement condition will be ignored in the following discussion.
Cognitive Response. H1 predicted that behavioral
commitment to avoid influenza vaccinations would be related to greater
depth of processing, such that non-vaccinators would report more total
cognitions than vaccinators. A t test showed no difference
between non-vaccinators and vaccinators (t (177)=.51, ns,
h=.03) (See Table 3).
Therefore, the analyses provided no support for H1.
H2 predicted that the dominant cognitive response produced
by non-vaccinators would be more negative than those produced by
vaccinators. A t test revealed a significant difference between
the groups (t (177) =2.282, p<.05,
h=.17), such that
non-vaccinators produced more negative dominant cognitive responses
than vaccinators. Thus, the analyses provided support for H2.
Biases in Message Processing. H3 predicted that
non-vaccinators would produce biased perceptions of the message by
underestimating one or more of the elements of the threat appeal. A
series of t test were analyzed for each of the elements:
severity, likelihood, response-, and self-efficacy. The t tests
revealed no significant differences between non-vaccinators and
vaccinators on severity (t (177)=.79, ns,
h=.06), likelihood (t
(177)=43.26, ns, h=.00),
or self-efficacy (t (177)=53.30, ns,
h=.05) (See Table 3).
However, analyses did reveal that non-vaccinators produced
significantly lower estimates of response-efficacy than did
vaccinators (t (177)=2.10, p<.05,
h=.16). Thus, the analyses
provided only limited support for H3.
Emotional responses to the persuasive message.
First, in order to address questions concerning emotional responses to
the threat appeal, we conducted a series of analyses to discern
whether or not each group realized a significant change in emotion
from baseline measures of each emotion to post-threat measures. This
analysis informs the reader as to whether or not the threat appeal
resulted in a change in emotion experience for the non-vaccinators and
vaccinators. For non-vaccinators, paired sample t tests
revealed that post-threat appeal measures of fear (t
(177)=11.83, p<.05), sadness (t (177)=3.32, p<.05),
and guilt (t (177)=3.19, p<.05) were significantly
higher than baseline measures, while post-threat measures of anger
were not significantly different from baseline measures (t
(177)=1.91, p=.06) (See Table 4). Among vaccinators, the only
emotion resulting in a significant change from baseline to post-threat
measures was fear (t (177)=10.33, p<.05). Analyses for
anger, sadness, and guilt (t (177)=.94, 1.03, .71 respectively)
failed to achieve statistical significance for the vaccinator group.
Second, for each emotion, we compared the change scores for emotional
experiences between the groups. H4 predicted that, due to the nature
of the persuasive message (i.e., a threat appeal), both vaccinators
and non-vaccinators would experience an increase in fear following
exposure to the message. A t test revealed no significant
differences between the groups in change in fear (t (177)=1.25,
ns, h=.10). Thus, the
analyses revealed an increase in fear for both non-vaccinators and
vaccinators, but the mean change scores for the groups were not
significantly different providing support for H4.
H5 predicted non-vaccinators would report a greater
experienced level of anger than vaccinators. A t test did not
show a significant difference in levels of anger change produced by
the groups (t (177)=.68, ns;
h=.06). Although the mean
change scores were in the predicted direction, the means were not
significantly different. Thus, H5 was not supported.
RQ1 sought to uncover the relations
between the experience of sadness and guilt following exposure to the
threat component of the message for vaccinators and non-vaccinators. A
t test was conducted to test for differences in emotion change
scores for sadness and guilt between the groups. The t test
revealed no significant differences between non-vaccinators and
vaccinators with regard to change in sadness (t (177)=1.52,
ns, h=.11). With regard
to guilt change, a t test did show a significant difference
between the groups (t (177)=2.65, p<.05,
h=.20), such that
non-vaccinators reported a significantly more guilt following exposure
to the threat appeal than did vaccinators.
Information integration. H6 predicted that
behavioral commitment to avoid influenza vaccinators would serve to
lower the magnitude of the correlation between dominant cognitive
response and attitude, such that non-vaccinators would produce a
correlation lower in magnitude than that produced by vaccinators. A
bivariate correlation revealed correlations of .48 (p<.05) and
.38 (p<.05) for non-vaccinators and vaccinators respectively
(See Table 5). A post-hoc z test was then conducted to test for
differences between the correlations, but the z-test did not
reveal a significant difference (z=.81, ns).
Furthermore, the pattern of the correlations ran counter to those
predicted by H6. Therefore, analyses provided no support for H6.
H7 predicted a relation between behavior commitment to avoid influenza
vaccinations and attitude certainty such that non-vaccinators would
report higher levels of attitude ambivalence than vaccinators. A t
test revealed a significant difference in attitude ambivalence between
the groups (t (177)=2.55, p<.05,
h=.19), such that
non-vaccinators reported higher attitude ambivalence reports than did
vaccinators (See Table 3). Therefore, the findings supported H7.
Bridging the gap from attitudes to behaviors. H8 predicted that
due to their behavioral commitment to avoid vaccinations,
non-vaccinators would report correlations between attitude and
behavioral intention lower in magnitude than vaccinators following
exposure to the message. A bivariate correlation revealed correlations
of .80 (p<.05) and .69 (p<.05) for non-vaccinators and
vaccinators respectively. A z test was then conducted to test
for differences, but none were revealed (z=1.65, p=.10).
Furthermore, the pattern of the correlations ran counter to those
predicted by H8 revealing no support for H8.
Finally, H9 predicted that behavioral commitment to avoid influenza
vaccinators would serve to lower the magnitude of the correlation
between behavioral intention and behavior, such that non-vaccinators
would produce a correlation lower in magnitude than that produced by
vaccinators. A bivariate correlation revealed correlations of .03 (ns)
and .24 (p<.05) for non-vaccinators and vaccinators respectively.
Again, a post-hoc z test of differences between the magnitudes
of the correlations was conducted but failed to identify any
significant differences (z=-.63, ns). Although the
pattern of the correlations was in the predicted direction, analyses
provided little support for H9.
There is a considerable body of evidence
that suggests threat appeals are effective in garnering attitude and
behavioral change; however, threat appeals are far from universally
effective. The findings here suggest that multi-motive processing
arising from prior behavioral commitments may explain some of the
limitations in the effectiveness of threat appeals.
Under single-motive processing, processors produce cognitions in an
attempt to develop or maintain an attitude either in favor of or
opposed to an advocated position. This processing leads to attitude
certainty, which is an antecedent of attitude-behavior consistency
(Gross et al., 1995). On the other hand, multi-motive processors
orient toward both cognitions in favor of and opposed to a
message advocacy. We argue that these processing demands lead to
attitude uncertainty or ambivalence and have implications for the
consistency between attitudes and behaviors. Dillard and Anderson
(2000) found that one of the strongest indicators of multi-motive
processing was attitude ambivalence (h=.46)
among non-, low, and high-tanners exposed to a threat appeal
concerning sun exposure. Attitude ambivalence reflects the tension
between the inherent incompatibility of competing motives to process.
In a sense, attitude ambivalence is a measure of one’s struggle to
integrate information from a persuasive message with attitudes held
prior to message exposure that indicts the held attitude.
was predicted that participants committed to avoiding flu
vaccinations, non-vaccinators, would report greater attitude
ambivalence than vaccinators. As predicted, non-vaccinators did report
greater attitude ambivalence than vaccinators. We argue that the
heightened ambivalence is derived from the tension between goals for
self-preservation and behavioral commitments faced under multi-motive
processing. Of note, the effect size for the finding was only
h=.19. Thus, the differences
between single- and multi-motive processing may not be as evident
given the issue at hand.
Prior research (Chaiken et al., 1989; Giner-Sorolla & Chaiken, 1997)
lead to predictions that multi-motive processors would produce more
total cognitions than single-motive processors. However, the additive
effect of validity seeking and defense motives was not realized in our
analyses. Instead, the data suggest a possible ceiling effect on the
cognitive capacity of participants based on the relatively large
number of cognitions produced by all participants (See Table 3).
Dillard and Anderson (2000) found a similar effect for depth of
processing among single- and multi-motive processors. Of note, both
studies obtained depth of processing measures with large standard
deviations. Further research controlling for individual differences
(i.e., need for cognition) may better inform this prediction. While it
is important to consider the volume of cognitive responses produced by
the groups, it is also informative to explore the valence of reported
cognitions. We discuss this issue below.
Based on the notion that defensive processing requires that one
counter-argue against a persuasive message and research that suggests
a relation between defense motives and negative cognitions (Giner-Sorolla
& Chaiken, 1997), we predicted that multi-motive processors would
produce cognitive reports more negatively valenced than would
single-motive processors. Analyses revealed that multi-motive
processors produced reports that were more nearly balanced in terms of
total supportive and counter-argumentative reports than were reports
by single-motive processors whose reports were significantly more
supportive in nature (See Table 3). For multi-motive processors in
this context, we believe that this balance between the reports
reflects the conflict faced by multi-motive processors, as they
attempt avoid the threat of contracting influenza and maintain
behavioral commitments to avoid vaccinations. Given some evidence of
single- and multi-motive processing, we now consider the potential
biases resulting from multi-motive processing in comparison to
had predicted that the tension multi-motive processors realize, as
they attempt to juggle conflicting goals, would result in biased
processing of the persuasive message. Specifically, these biases would
take the form of systematic underestimations of the various elements
of the threat appeal—likelihood of the threat, severity of the threat,
response efficacy of the recommended actions, or self efficacy to
carry out the recommended actions. Analyses revealed that the only
significant between groups difference was obtained for the response
efficacy variable. Consistent with our prediction, non-vaccinators
produced a lower estimate for the efficacy of the recommendations than
did vaccinators. The pattern of the findings lends some support to two
conclusions. First, both vaccinators and non-vaccinators exhibit
indicators of validity processing. The largely similar means for
likelihood, severity, and self efficacy indicate similar perceptual
processes, as participants attempt to validate the threat appeal.
Second, the underestimation of response efficacy by non-vaccinators
may reflect an attempt to rectify the conflict between the concurrent,
competing motivations to maintain one’s behavioral commitment and
simultaneously resolve a threat to self-preservation. In other words,
non-vaccinators underestimate the efficacy of getting an influenza
vaccination in order to justify inattention to the threat and
maintaining behavioral commitments to avoid vaccinations. While this
bias is informative, we also considered how affective experiences
might inform the findings.
responses to the persuasive message
Consistent with our predictions, both vaccinators and non-vaccinators
experienced fear arousal following exposure to the threat appeal, and
the level of arousal experienced was not significantly different
between the groups. This arousal suggests that both groups recognize
influenza as a threat to self-preservation goals and therefore process
the message in an effort to validate attitudes and behaviors
concerning the threat. The findings provide some evidence of
motivation toward validity seeking processing by both groups, not just
With regard to the experience of other
emotions, vaccinators failed to experience a significant change in
emotion for baseline to post-threat measures of anger, sadness, and
guilt. On the other hand, non-vaccinators experienced a significant
change from baseline to post-threat in sadness and guilt (of note,
anger change from baseline to post-threat approached significance).
Sadness occurs when one appraises the environment and discerns a loss
of well-being which one is unable to restore (Frijda, 1986; Lazarus,
1991). For non-vaccinators, appraisal revealed a loss and resignation
resulting in experiences of sadness due to the inherent
incompatibility of their competing goals. These findings appear to be
consistent with multi-motive processing demands.
Similarly, non-vaccinators also
experienced guilt which is experienced when one appraises that they
have failed to meet some personal standard (i.e., maintaining my
commitment to a valued behavior) (Frijda, 1986; Lazarus, 1991). In
this study, it appears that multi-motive processors recognized the
likelihood, the severity, and their ability to carry out threat
averting recommendations, yet their behavioral commitment to avoid
vaccinations compelled them to forego carrying out the
recommendations. It appears that one method of relieving the tension
of multi-motive processing is to underestimate the efficacy of the
recommended response. However, for multi-motive processors, this
practice results in the arousal of guilt, which is experienced as one
recognizes his or her failure to attend to self-preservation goals.
A final note with regard to anger is also
warranted. While levels of statistical significance were not realized
for both baseline to post-threat measures of change and between groups
change in anger, the pattern of the means was in the predicted
direction. While it appears that non-vaccinators realized the
implications of the persuasive message with regard to their loss of
autonomy, sadness and guilt experiences appear to be better
illustrators of defensive processing. This finding in combination with
an increase in fear as an indication of validity processing lends
further support to the presence of multi-motive processing and its
Integration: From Message Processing to Behavior
success of persuasive messages is contingent upon the integration of
cognitions about the message into attitudes, and subsequently,
transforming attitudes into actual behaviors. This process of
integration and transformation can be assessed through three distinct
steps: (a) integrating information into attitudes, (b) transforming
attitudes into plans to carry out the behavior, and (c) transforming
plans into actual behavior. At the first step, persuasion requires
that the message processors engage the information (e.g., evidence,
arguments, etc.) presented in the threat appeal and report attitudes
toward the advocacy consistent with the cognitions. We had predicted
(H6) that multi-motive processors, due to the conflicting nature of
their motivations, would not successfully integrate information
presented in the threat appeal into attitudes toward the message. The
findings did not support the prediction. Consistent with validity
seeking processing, both vaccinators and non-vaccinators produced
significant positive correlations between dominant cognitive response
and attitude reports. However, as suggested earlier, this successful
integration does not insure persuasion.
the second step, persuasion relies upon the transformation of
attitudes toward the message advocacy into plans to carry out a
behavior (i.e., behavioral intention). Again, we had predicted smaller
correlations between attitude reports and behavioral intention for
multi-motive processors; however, the results indicated significant
positive correlations between the variables of interest for both
vaccinators and non-vaccinators. While the correlations do suggest
strong evidence of validity seeking orientations for both groups,
behavioral intentions must still be transformed into behavioral
actions before persuasion can be successful.
the final step, message processors must transform plans to carry out
the behavior into actual behaviors. Here, we had predicted that the
conflicted nature of multi-motive processing would result in
significantly lower correlations between behavioral intention and
behavior as compared to single-motive, validity processors. For
vaccinators, the analyses reveal a significant relationship between
behavioral intention and behavior. With significant correlations at
each step of the persuasion process, it appears that vaccinators are
able to successfully integrate the information offered by the threat
appeal and transform that information into a preventative health
behavior. For non-vaccinators, this process is not realized, and we
begin to see the influence of multi-motive processing evidenced at
earlier stages of cognitive and affective processing. Non-vaccinators
failed to transform their behavioral intentions into behaviors, as
evidenced by the lack of a significant correlation between behavioral
intention and behavior. However, this evidence is tempered by the lack
of a significant difference between the correlations produced by
vaccinators and non-vaccinators. We had anticipated that multi-motive
processing would impact the integration of information and the
transformation of attitudes into behavioral intention, but it appears
that the multi-motive processing’s impact on the persuasion process
occurs at the point which behavioral intentions are transformed into
an actual health behavior. For non-vaccinators, despite moderate to
strong relationships between dominant cognitive responses and
attitudes, as well as attitudes and behavioral intention, they still
fail to transform these reports into actual behaviors. This finding is
not uncommon in the literature where we find that the correlation
between behavioral intention and behavior is less than perfect (Fishbein
& Azjen, 1975). The findings here lend some support to multi-motive
processing as an explanation for these findings. It may be that
behavioral intentions reported by message processors reflect evidence
of validity processing, yet these intentions fail to get carried out
due to conflict arising from concurrent defense motive processing in
an effort to meet some other goal salient to the issue.
general, our predictions here failed to receive support. However, the
findings clearly indicate evidence of validity motive processing by
both vaccinators and non-vaccinators. Furthermore, multi-motive
processing may explain the lack of a significant relation between
behavioral intention and behavior for non-vaccinators as the tension
of concurrent defensive and validity seeking processing demands is
realized at this point in the persuasion process. Thus, multi-motive
processing may explain the limited effectiveness of persuasive message
in garnering behavioral change. These findings suggest a series of
implications for researchers and applied health researchers.
I. (1991). The theory of planned behavior. Organizational Behavior
and Human Decision Processes, 50, 179-211.
American Medical Association. (1999). Do you have the flu? The
Journal of the American Medical Association, 281, 962.
Anderson, N. H. (1971). Integration theory and attitude change.
Psychological Review, 78, 171-206.
F. J., & Mongeau, P. (1984). Fear-arousing persuasive messages. In R.
Bostrum (Ed.)., Communication yearbook, vol. 8 (pp. 330-375).
Newbury Park, CA: Sage.
S.S., & Brehm, J.W. (1981). Psychological reactance: A theory of
freedom and control. New York: Academic Press.Brehm, (1984).
for Disease Control (August, 1999). http://www.cdc.gov.
Chaiken, S., Liberman, A., & Eagly, A. H. (1989). Heuristic and
systematic information processing within and beyond the persuasion
context. In J.S. Uleman & J.A. Bargh (Eds.), Unintended thought,
(pp. 212-252). New York: Guilford Press.
Chapman, G. B., & Coups, E. J. (1999). Predictors of influenza vaccine
acceptance among healthy adults. Preventive Medicine: An
International Journal Devoted to Practice and Theory, l29, 249-262.
Dillard, J.P. (1994). Rethinking the study of fear appeals: An emotional
perspective. Communication Theory, 4, 295-323.
Dillard, J. P., & Anderson, J. W. (2000). Single-motive and
multi-motive processing of a threat appeal: Promoting a healthy
orientation to sun exposure. Unpublished manuscript.
Dillard, J.P., Kinney, T.A., & Cruz, M.G. (1996). Influence, appraisals,
and emotions in close relationships. Communication Monographs, 63,
Dillard, J.P., & Peck, E. (2000). Affect and persuasion: Emotional
responses to public service announcements. Manuscript under review.
Dillard, J. P., Plotnick, C. A., Godbold, L. C., Freimuth, V. S., &
Edgar, T. (1996). The multiple affective outcomes of AIDS PSAs: Fear
appeals do more than scare people. Communication Research, 23,
Festinger, L. (1957). A theory of cognitive dissonance. Stanford,
CA: Stanford University.
Fishbein, M., & Azjen, I. (1975). Belief, attitude, intention, and
behavior: An introduction to theory and research. Reading, MA:
N.H. (1986). The emotions. Cambridge, UK: Cambridge University
Giner-Sorolla, R., & Chaiken, S. (1997). Selective use of heuristic and
systematic processing under defense motivation. Personality and
Social Psychology Bulletin, 23, 84-97.
S. R., Holtz, R., & Miller, N. (1995). Attitude certainty. In R. E.
Petty & J. A. Krosnick (Eds.), Attitude strength: Antecedents and
consequences. Mahwah, NJ: Lawrence Erlbaum Associates.
G.M. (1986). The instrumentality of attitudes: Toward a neofunctional
theory. Journal of Social Issues, 42, 99-114.
Herman, C. J., Speroff, T., & Cebul, R. D. (1994). Improving compliance
with immunization in the older adult: Results of a randomized cohort
study. Journal of the American Geriatrics Society, 42, 1154-1159.
Katz, D. (1960). The functional approach to the study of attitudes.
Public Opinion Quarterly, 24, 163-204.
M. S., & Hunter, J. (1993a). Attitude-behavior relations: A
meta-analysis of attitudinal relevance and topic. Journal of
Communication, 43, 101-142.
M. S., & Hunter, J. (1993b). Relationships among attitudes, behavioral
intentions, and behavior: A meta-analysis of past research, Part 2.
Communication Research, 20, 331-364.
Lazarus, R. (1991). Emotion and adaptation (pp. 42-126). New
York: Oxford University Press.
Leventhal, H. (1971). Fear appeals and persuasion: The differentiation
of a motivational construct. American Journal of Public Health, 61,
Mongeau, P. A. (1998). Another look at fear-arousing persuasive appeals.
In M. Allen & R. Preiss (Eds.), Persuasion: Advances through
meta-analysis (pp. 53-68). Cresskill, NJ: Hampton Press.
Morley, D. D. (1987). Subjective message constructs: A theory of
persuasion. Communication Monographs, 54, 183-203.
S. E., Furumoto-Dawson, A., Monto, A. S., & Fasano,N. (1995). Influenza
vaccine use among an elderly population in a community intervention.
American Journal of Preventive Medicine, 11, 271-276.
Rogers, R.W. (1975). A protection motivation theory of fear appeals and
attitude change. Journal of Psychology, 91, 93-114.
Shaver, P., Schwartz, J., Kirson, D., & O’Connor, C. (1987). Emotion
knowledge: Further exploration of a prototype approach. Journal of
Personality and Social Psychology, 52, 1061-1086.
M.B., Bruner, J.S., & White, R.W. (1956). Opinions and personality.
New York, NY: Wiley.
University of Wisconsin Hospitals and Clinics (1999). Influenza and
influenza vaccine. Health Facts for You. Produced by the
Department of Nursing.
Witte, K. (1994). Fear control and danger control: A test of the
K. (1993). Message and conceptual confounds in fear appeals: The role
of threat, fear, and efficacy. The Southern Communication Journal,
Health Organization, (1999). Who tags 1999-2000 vaccine, prepares flu
pandemic plan? Public Health Reports, 114, 211.