\HIWHHHWHW

!

 

a WIN“lWl\\\|fll|Ull|\\WW“

936

LIBRARY
Michigan State
University

This is to certify that the
thesis entitled

An Examination of the Motivational Effect of Anxiety on
Persuasive Message Processing

presented by

Sarah Katherine Foregger

has been accepted towards fulfillment
of the requirements for the

Master of Arts degree in Communication

 

 

' Major Professor’s §ignature

0mm / 1. 4007
Date

MSU is an Affinnative Action/Equal Opportunity Institution

 

PLACE IN RETURN Box to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2/05 mfiatmjndd-pjs

 

AN EXAMINATION OF THE MOTIVATIONAL EFFECT OF ANXIETY ON
PERSUASIV E MESSAGE PROCESSING

By

Sarah Katherine Foregger

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
MASTER OF ARTS

Department of Communication

2004

ABSTRACT

AN EXAMINATION OF THE MOTIVATIONAL EFFECT OF ANXIETY ON
PERSUASIVE MESSAGE PROCESSING

By

Sarah Katherine F oregger

This study examines the effect of anxiety on type of persuasive message
processing route. It was predicted that increased levels of anxiety would relate to
increased systematic processing. In turn, increased systematic processing was predicted
to be related to increased memory performance. Fifty-eight students from
communication courses at a large Midwestern university participated voluntarily; they
read a persuasive message and in order to determine systematic processing, listed their
thoughts about, and completed a memory measure for the message. Their anxiety levels
were assessed during different points of the study using the state portion of Spielberger’s
STAI. Results from this study show that anxiety did not relate significantly to systematic
processing. The Pearson correlation between anxiety and systematic thoughts was
negative, and non-significant, r = -.176, p = .187. It was also postulated that as
systematic thoughts increased, so too would memory performance as measured by “hits.”
The Pearson correlation coefficient between hits and systematic thoughts was both
negative and non-significant, r = -.l 19, n.s. Although non-significant, results indicate
that since anxiety does not function in this study as the Negative State Relief Model
suggests, it deserves further study in relation to message processing routes, as well as

consideration as to whether it is a negative or positive state.

DEDICATION PAGE

This thesis is dedicated to my parents, Mary P. and Joseph H. Foregger, who instilled in
me the value of education and learning, and for whom I continue on this path.

I also remember Oliver, who I never made it home for.

iii

ACKNOWLEDGEMENTS

First and foremost, I would like to thank my advisor, Dr. Sandi W. Smith, for her
encouragement in all things during my masters program. Her assistance and belief in me
has been invaluable. From our first meeting at the annual department picnic, where we
bonded over Williams-Sonoma bake-ware, I’ve admired her as both a scholar and person.
I would also like to acknowledge my committee members, Dr. Franklin Boster and Dr.
Chuck Salmon, for their insight and input during my first foray into experimental
research. I feel lucky that they agreed to lend their time and expertise to my project.

Additionally, I would like to thank the people in my life whose support and
influence have brought me to this point: my mom and dad, for supporting me throughout
my educational pursuits; my grandmother, Eleanor Erdevig, for setting an admirable
example of educational achievement; and my aunt, Susan Willemsen, for her
unconditional love, acceptance, and timeless advice.

I would not be at Michigan State University’s Department of Communication
were it not for Dr. Gary Meyer, my persuasion professor at Marquette University, who
believed that there was “no other place to study persuasion and play football on Fridays,
but MSU.” He was right about both.

Over the course of two years at MSU, I’d like to think I’d have made it on my
own. But as Joe Cocker sang, we truly do “get by with a little help from our fi'iends.” I
want to thank those who have helped me to “get by,” whether they were aware of it or
not: my ‘dyad’ Cat, my master’s cohort -— Ed, Jeffrey, Al; my of wonderful R.A.’s
(especially Nat and Nick), Carrie Bednar, my best friend since our Olympic-running

days; Janelle & Scott, Finch, my guide on the CFA adventure; Katie & Amber for

iv

lunchtimes, continuous US magazine supply, and much-needed trips to New York; Alina
for providing a quiet refuge, my brother Daniel, for being there; and finally, my cousin
Laura for dancing with me to Michael Jackson.

Finally, I would like to thank the Dept. of Residence Life for funding my master’s
degree, in exchange for a piece of my soul (not to mention my sanity). I will also be
eternally grateful to MSU’s “54-day-rule” regarding graduate assistantships for the

restoration of my life.

TABLE OF CONTENTS

LIST OF TABLES ................................................................................... vi
INTRODUCTION .................................................................................. vii
Anxiety ........................................................................................ 2
Negative State Relief Model ............................................................... 3
Elaboration Likelihood Model ............................................................. 6
Heuristic — Systematic Model .............................................................. 7
Affect and Dual Process Models ........................................................... 9
Rationale .................................................................................... ll
Hypotheses .................................................................................. 11
METHOD ............................................................................................ 12
Overview .................................................................................... 12
Sample ....................................................................................... 12
Procedure .................................................................................... 13
Induction .................................................................................... 13
Measures .................................................................................... 14
Anxiety Measure ........................................................................... 14
Positive Feeling Measure .................................................................. 14
Thought Listing ............................................................................. 15
Memory Measure ........................................................................... 16
RESULTS
Induction Check ............................................................................ l7
STAI .......................................................................................... 17
Hypothesis One ............................................................................. 19
Hypothesis Two ............................................................................ 20
DISCUSSION
The Effect of Anxiety on Systematic Processing ...................................... 21
Behavioral Intention and Memory Performance ....................................... 22
Limitations .................................................................................. 23
Summation .................................................................................. 25
REFERENCES
APPENDICES

Appendix A: College Math Skills Test
Appendix B: Persuasive Messages
Appendix C: Memory Measure
Appendix D: Footnotes

vi

LIST OF TABLES

Table 1 Means, Standard Deviations, and Ranges for all Variables .................... 21
Table 2 Mean Table of Behavioral Intention, Criterion Bias, and Sensitivity. . . . . ....23
Table 3 Mean Table of Message Type, Criterion Bias, and Sensitivity ................ 24

vii

INTRODUCTION

Many studies have examined the impact of negative and positive affect on the
persuasive message-processing route used in the Elaboration Likelihood Model (ELM) or
the Heuristic-Systematic Model (HSM). However, in regard to anxiety and processing
routes, extensive work has yet to be done. From some previous findings, it would seem
that those who are anxious should have diminished cognitive capacity for new
information processing as the basic cognitive symptoms of anxiety include difficulty
concentrating, narrowed attention, and distorted reasoning (Perez-Lopez & Woody,
2001), which imply impaired cognitive processing. Findings in mood research, however,
suggest the opposite - that those in negative moods, such as feeling sad, angry, or guilty
(Mitchell, 2000; 2001), tend to process messages more carefully and systematically
compared to their positive mood counterparts, who tend to rely on heuristic cues (Bless,
Mackie, & Schwarz, 1992; Mackie & Worth, 1989, 1991; Schwarz & Bless, 1991; Wells,
1999). Findings also indicate that negative mood states prompt systematic processing
and positive mood states prompt heuristic processing (Mackie & Worth, 1989, 1991;
Schwartz & Bless, 1991). As anxiety is a negative affect (Russell & Pratt, 1980; Wells,
1999), it seems contrary to the information presented on cognitive symptoms of anxiety
to find that those in an anxious state are more likely to process systematically than those
in a happy or joyful mood. As such, a test of the way in which anxiety functions within
the tenets of the HSM is necessary.

A variety of research in the area of information processing has touched on the
function of anxiety in areas such as message recall, bias, attention, retention, and

persuasion (Ruiz-Caballero & Bermudez, 1997; Weary & Edwards, 1994; Wells, 1999).

However, research needs to be done on the effects of anxiety in regard to the predictions
of choice of message processing route suggested by the dual process persuasion models
of the ELM and the HSM. Therefore, this paper examines how anxiety may effect
information processing of persuasive messages. The HSM is used to determine if
anxious individuals are more likely to process messages through a systematic route.

Taking the findings on affect and persuasive message processing route into
consideration, this paper proposes that anxiety serves as a motivation to process
systematically using the rationale from Cialdini’s Negative State Relief Model (N SRM)
(1973). Cialdini’s NSRM is used as an explanation to explain why those who are anxious
may cognitively process messages systematically despite research findings that anxiety
should impair cognitive functions and ability. The proportion of the persuasive message
recalled correctly by anxious respondents as well as the number of systematic thoughts
created afier exposure to the message will be measured and correlated with anxiety
scores. Within the literature review, dual process models of persuasion, the ELM and the
HSM will be outlined, previous research regarding affect and persuasion will be
reviewed, a brief description of anxiety will be presented and the NSRM will be
explained. Finally, the relationship between anxiety, systematic processing in dual-
processing models, and the propositions of the NSRM will be discussed.

The main proposition here is that anxiety, as a negative affect serves to motivate
individuals to attend to and process messages through a systematic route, based on the
predictions of the NSRM. As such, those high in anxiety should be significantly more
apt to process positive feeling persuasive messages through systematic routes and

elaborate on those messages than their less anxious counterparts.

Anxiety

Anxiety is an unpleasant emotional state characterized by symptoms of muscle
tension, worry, restlessness, and uneasiness that ofien require cognitive effort and energy
to manage (Wells, 1999). In past research, anxiety and fear have been used
interchangeably, when in reality they are two separate concepts (Eyseneck, 1997;
Goodwin, 1986). According to Goodwin (1986), fear is an emotion with an identified
danger, whereas in anxiety the source of distress is unknown, with individuals feeling a
comparatively unjustifiable intensity of emotion.

While the term “anxiety” has come to encompass a vast array of physical and
mental symptoms and spans across situations, anxiety remains one of the main reasons
people seek psychological help (Oei, Moylan, & Evans, 1991). The prominence of
anxiety among members of our society warrants delving into the potential implications
for it to affect information processing.

Despite its continued societal presence, during the past fifty years inconsistency
among researchers regarding use of the term “anxiety” has made definition and research
difficult. Many social scientists, including Spielberger (1966), Freud (1936), Goodwin
(1986), Cattell (1962), Scheier (1962), Neimah (1981), and Eysenck (1997) have
attempted to provide conceptual definitions of anxiety. Among the definitions of anxiety,
a common thread does exist. Anxiety definitions, despite other differences, either focus
directly upon, or contain, one or all three of the following aspects: a behavioral
component of anxiety, a subj ectivc component, and/or a physiological component. As
such, the following definition of anxiety, which is a mix of previous definitional

strengths, is presented: “anxiety is an unpleasant emotion with definite physical

sensations, linked with cognitive system functioning, involving a degree of heightened
arousal and or disproportionate mental preparation for some unforeseen future threat or
danger.”

Cattell (1962) and Scheier (1962) attempted to further quell confusion regarding
the use of anxiety in research through analyses of previously used anxiety variables.
They used factor analysis to determine that the roughly 800 variables indeed fell under a
single general factor of anxiety, supporting the notion that anxiety is one concept. Cattell
and Scheier both also found, however, that two distinct factors can be identified from the
variables, and labeled them according to their properties, state anxiety and trait anxiety.

The difference between state anxiety, trait anxiety, and anxiety as a general term
is an important, but frequently overlooked distinction in anxiety research. State anxiety
is an unpleasant emotion that varies both over time and intensity whereas trait anxiety is
an individual difference in anxiety proneness as a personality trait (Scheier, 1962).
Spielberger (1966, 1999) expanded upon Cattell and Scheier’s definitions and
discoveries, developing the often-used State-Trait Anxiety Inventory (STAI), a scale for
distinguishing between the two, and adding to the existing definitions. State anxiety, as
defined by Spielberger (1999) is a “temporal cross-section in the emotional life of a
person, consisting of subjective feelings of tension, apprehension, nervousness, worry,
and activation (arousal) of the automatic nervous system.” Spielberger discusses trait
anxiety in terms of being a relatively stable individual difference as was found in Cattell
and Scheier’s definition, but in addition, includes, “a disposition to respond to such
situations with more frequent and intense elevations in state anxiety.” The symptoms of

anxiety remain the same in both state and trait anxious people, and should both affect

information processing. Additionally, the proposed definition appropriately encompasses
anxiety as a general concept, yet can still contain state or trait anxiety.

For the purpose of this study, also it is important to note that anxiety should
function as a negative mood state. Russell and Pratt (1980) sought to determine the
affective adjectives connected with environmental perceptions and place them on
dimensions based on their affective quality. From their research, they determined a two-
dimensional affective quality space, with borders of unpleasant to pleasant and sleepy to
arousing. Distress was found to be one of the eight primary meanings conceptualized,
and was also found to be negatively valenced. Distinguished by adjectives such as
“panicky,” “frenzied,” and “tense” and related to the notions of psychological stress,
unpleasant, and arousing feeling; distressing affect quality appears extremely similar to
the affect of anxiety. Given the previous descriptions of anxiety, the presented
definitions, and Russell’s finding of distressing quality, it seems accurate to categorize
anxiety as a negative mood state.

Anxious individuals may use the systematic processing of messages as a form of
distraction from anxiety. Similar to Cialdini’s predictions in the NSRM, Schwartz and
Bless (1991) postulate that a negative state indicates that an action to relieve the
negative mood needs to occur and serves as a motivator for the assessment of their
environment for potential ways to alleviate their negative state.

Negative State Relief Model (NSRM)

Robert Cialdini (1973) proposed in the NSRM that a negative mood produces a

drive to alleviate negative feelings. Originally proposed as theory within the realm of

helping and altruistic behavior, Cialdini found that people in a negative mood are

motivated by the mood itself to behave in a way that will realign their mood to a more
positive or neutral point. NSRM postulated that people who felt bad upon watching
another harmed would engage in a helping action to alleviate their own bad feeling, and
the results of his study affinned this belief. While the original model was created to
explain altruism, Cialdini noted that the NSRM could be viewed in a broader sense,
explaining actions that occur by those in a negative state as ways in which they are
striving to reconcile their negative mood states (1973, 1976).

The negative mood state discussed in the NSRM can be conceptualized very
broadly, thus it can contain any negative affect. Therefore according to the NSRM,
anxious individuals should be more apt to scrutinize messages for a source of positive
affect, because they are motivated to relieve their mood state and as such, search their
environment for clues about how to do so. In the NSRM, the mood state functions as the
motivator that causes increased attention and focus to environmental stimuli as both a
means of distraction and also to determine if the message content would be useful in
achieving the anxious subject’s goal mood state of relief. From the NSRM it could be
inferred that those in a negative mood are more likely to process systematically because
they are motivated to act in some way to attempt to relieve their own mood state.
Elaboration Likelihood Model

Petty and Cacioppo’s (1986) Elaboration Likelihood Model explains how
receivers process persuasive messages. At its core, the ELM is founded on the
assumption that people are “cognitive misers,” and as such, choose to attend to messages
and allocate resources based on their purpose for message attention and processing. In

the model, a central and a peripheral route of message processing are proposed. The

central route is defined by effortful cognitions, requiring the receiver to process the
incoming knowledge in terms of what may already be known, creating links and
elaborating on information presented. In contrast, processing that occurs via the
peripheral route relies more on simple heuristic cues given in the message or by the
sender and conserves cognitive energy (Petty et al., 1994).

A receiver’s tendency towards central or peripheral processing after receiving a
message is often determined by two factors: whether they have the motivation to
elaborate on the issue and whether they have the ability to elaborate on the issue. Studies
have found that ability to elaborate on the issue is often influenced by the presence of
distractions in the persuasive setting and the receiver’s prior knowledge about the topic
(Slater, 2000). Receiver’s motivation has been found to be influenced by several
elements, including personal need for cognition level and personal relevance of the topic
to the receiver (Petty et al., 1994). Low levels of motivation or ability result in a greater
likelihood of peripheral processing. The ELM allows for several processing goals,
however, none of which contain emotional state relief as a potential motivator to process
incoming messages. Similar to the ELM, the Heuristic-Systematic Model concerns
receivers’ message processing routes.

Heuristic-Systematic Model

The Heuristic-Systematic Model (HSM), like the ELM, is a dual process model
with a systematic and heuristic route to persuasion. Receivers may process messages
systematically, engaging in more analytical thought, elaboration, and judgment of
messages; or they may process heuristically, directing focus to exigent message cues that

trigger heuristics to decipher meaning and form attitudes, or they may combine the two,

as they are not seen as mutually exclusive in this model (Todorov et al., 2002). Similar to
the ELM, the systematic route requires more cognitive effort, whereas the heuristic route
does not.

The HSM and the ELM, dual process cognitive-response models of persuasion
often treated in literature as nearly identical, are models that encourage a persuader to
take into consideration the mental capacity of the receiver to process the given message.
According to the HSM, “people engage in systematic processing of persuasive
information only when they are sufficiently motivated. . .however, if they are not
sufficiently motivated or do not have sufficient cognitive resources, they can engage in
superficial or heuristic processing” (Todorov, Chaiken, & Henderson, 2002, p. 196).
While either route or an additive effect of both routes can lead to attitude change, it is
generally accepted that attitudes changed through careful elaboration and systematic
analysis of the presented message are stronger than weaker attitudes formed through use
of heuristic cues (Mitchell, 2000). As such, many persuaders aim to have their audience
process systematically as a way of creating enduring attitudes.

The HSM, like the ELM, also identifies two qualifications, motivation and ability,
that determine persuasive message processing route used by the receiver to process
incoming messages. Motivation in the HSM is defined slightly differently than in the
ELM, encompassing a qualitative and quantitative dimension. It is proposed that the
quantitative assumption conceives motivation as a function of the discrepancy between
the receiver’s actual confidence and desired confidence for a task. With more
discrepancy in the message confidence level comes a greater propensity to engage in

systematic processing. Three qualitative motivations for processing information have

been discovered. These qualitative motivations state that receivers may be internally
motivated by accuracy, defense, or impression to process information via either a
heuristic or systematic route or both (Todorov et al., 2002). If a receiver is sufficiently
motivated and possesses the ability, messages will be processed via the systematic route,
which, congruent with the central route in the ELM, produces attitudes more resistant to
counterargument. What has yet to be studied in depth is whether a mood state can
function as a motivating factor for message elaboration, specifically, whether anxiety, a
negative state, could firnction within the ELM or HSM as a motivator for attention and
elaboration.
A fleet and Dual Process Models

The dual process models have important implications for attitude formation and
persuasive argument judgment. As such, discovering the variables that may influence the
choice between cognitive processing routes would be valuable knowledge in the field of
persuasion. Receiver’s mood states are one variable that have been studied to determine
their possible influence on processing route. The affective state of the receiver is
important to consider, because according to Nabi (2002), affect can stimulate careful
information processing and can direct the depth or path of information processing.

Several studies have been conducted to determine the possible mediating effect of
emotion on cognitive processing and persuasive messages (Bless et al., 1992; Bless,
Schwarz, & Strack, 1990; Mackie & Worth, 1989, 1991; M. Mitchell, 2000, 2001; Petty
et al., 1994; Schwarz & Bless, 1991). When presented with either strong or weak
persuasive messages, the majority of studies conducted have found receivers in a positive

mood to be less likely to elaborate on messages, less likely to process systematically, and

more likely to rely on heuristics for judgment (Bless et al., 1992; Bless et al., 1990;
Mackie & Worth, 1989, 1991; Schwarz & Bless, 1991). In contrast, those in negative
affective states were found to be more analytical, more likely to use logical reasoning
when presented with a message, pay closer attention to the message as judged by
message recall (Mackie & Worth, 1991) and be more apt to elaborate on messages as
judged by message relevant thoughts (Bless etal., 1992; Schwarz & Bless, 1991). In
addition, those in a negative state were found to have a narrowed focus on the message
(Schwarz & Bless, 1991) as compared to those in a positive state, creating more message
relevant connections, as well as processing the messages more systematically overall
(Mackie & Worth, 1989, 1991; Schwarz & Bless, 1991). Mackie and Worth (1989) did
find significant use of the systematic route by those in a negative affect state, however
they also found that receivers in a positive state, if given longer amounts of processing
time, were eventually able to elaborate more on the message.

Mood states have not been found to alter cognitive processing route preferences in
all studies. Mitchell (2000, 2001) did not find a significant effect for positive mood
versus negative mood regarding systematic or heuristic routes, however she did find that
respondents in happy, sad, and angry mood states all processed the persuasive messages
differently, indicating that different moods do in fact cause receivers to utilize different
processing strategies.

Schwartz and Bless (1991) as well as Weary and Edwards (1994) have both
postulated that a negative mood serves to motivate the receiver to assess the environment
for cues alerting danger. This heightened sensitivity to emotional cues may cause

receivers in a negative state to focus their attention on messages within their

10

environment. Anxious individuals already have a higher arousal level than their non-
anxious counterparts and, according to Weary and Edwards (1994), are focused on
environmental cues of impending negative events.

These varied findings in regard to message processing and mood in relation to the
HSM model can best be explained if combined with action explanations found in the
NSRM. Thus, this study postulated that the negative mood of anxiety is a motivational
factor that leads to use of the systematic route of information processing.

Rationale

According to the NSRM, people are motivated by negative mood states to find
relief. Perhaps this drive for emotional balance focuses the attention of the receiver on
messages in the environment, allowing for more systematic processing. The proposal
here, based on Cialdini’s NSRM, is that the negative state of anxiety drives the receiver
to attend to a message as it may contain potential relief from the negative state. As such,
this drive caused by anxiety serves as a sufficient motivator for an individual to allocate
cognitive resources toward careful information processing through systematic routes.

If positive affect decreases the probability of systematic processing, and those
who are anxious are likely to process more systematically, anxiety as a negative affect
must serve some other function if it allows for systematic processing. Thus, Cialdini’s
NSRM serves as explanation of the function of anxiety as a motivator in the persuasive
message processing paths proposed by the HSM. As such, the following hypotheses are
proposed:

Hypotheses

ll

H1: As anxiety increases, individuals will be more apt to process information in
messages systematically.
Given that systematic processing has been found to form cognitions that are longer
lasting (Petty et al., 1994; Slater, 2002) the following hypothesis is also proposed:

H2: As systematic processing, as assessed through the thought listing task,
increases so will “hit” rates on a recognition memory measure.

Method

Overview

In this study, participants were asked to report their anxiety levels using
Spielberger’s State-Trait Anxiety Scale. In order to ensure variance on the independent
variable, an anxiety induction was used and participants were selected at random for the
anxiety induction group. They were then presented with a persuasive message to visit
websites where clicking a button would either contribute to saving the rainforest or
feeding animals at shelters. The messages were identical in length, inclusion of statistics,
and information. Participants were then asked to complete a thought-listing measure,
designed to assess systematic/heuristic thought processes; a recognition memory test, and
completed the state portion of the STAI as a post-test measure of anxiety.
Sample

Fifty-eight students enrolled in undergraduate communication courses at a large
Midwestern university participated in this study. Students received extra credit for their
research participation.

Procedure

12

Participants were randomly selected to either receive the state anxiety induction
or participate without the induction. A state anxiety induction was used to ensure that
variance in anxiety would exist within the sample. To begin, both groups of participants
completed Spielberger’s State-Trait Anxiety Inventory (STAI), with the state portion
given first. In the anxiety induction group, the lab room also included a video camera
set-up, not included in the room for the control group. Participants who were selected to
receive the anxiety induction were then given a brief “College Math Skills Test,” and
told it would be corrected while they read the persuasive message. Immediately after
completion of the math test, the group took the state portion of the STAI as an induction
check. Both the control and the induction group were given a positive feeling persuasive
message, which encouraged them to visit websites where clicking a button would either
contribute to saving the rainforest or feeding animals at shelters. Assignment of message
topic was random across the control and induction groups. After reading the message,
participants completed a thought listing survey and proceeded to the recognition test of
memory, designed to determine the extent to which they recalled the positive feeling
message. Finally, the state portion of the STAI was given to assess anxiety levels.
Induction

Six groups each with five subjects were randomly selected to participate in an
anxiety induction designed to increase state anxiety levels. Research on previously used
anxiety inductions led to the creation of the induction used in this study (Blanchette &
Richards, 2003; Bright and Freedman, 1998; Hall & Crisp, 2003; Mogg, Kentish, &
Bradley, 1993). Subjects selected to receive the induction entered the same laboratory

room as used for the control subjects. In the comer of the room was a video camera,

13

which was not actually used to tape the sessions, but to create the perception of future
evaluation by others and increase anxiety. Additionally, a brief math test entitled
“College Math Skills Test” (Appendix A) was administered and subjects were told it
would be corrected during their time in the lab. After completing the math test, anxiety
levels were checked using the state portion of the STAI.
Measures
Anxiety Measure

Spielberger’s State-Trait Anxiety Inventory (STAI) was used to measure anxiety,
for later assessment of unexpected differences in state and trait anxiety functions on
recall. The STAI is one of the most commonly used inventories to assess anxiety and has
been used cross-culturally and in a variety of situations. The reliability (or between .85
and .95) for the STAI has been found to be acceptable for internal consistency (Barnes,
Harp, & Jung, 2002). The STAI has forty items, with twenty items comprising each the
state and trait portions. “Anxiety absent” items are reverse scored and the twenty items
on each scale are summed to compute the scale total. State and trait scale scores are then
added to provide overall STAI score. It is advised that the state scale be given prior to
the trait scale. Sample items from the state scale include “I feel fiightened” and “I feel
pleasant,” and sample items from the trait scale include “I wish I could be as happy as
others seem to be,” and “I have disturbing thoughts.”
Positive Feeling Message

One of two positive feeling messages‘ were given to participants to read, both
encouraging visiting a website where clicking a button supports a charitable cause, either

saving the rainforest or giving food to animal shelters. Two messages were used to

14

eliminate the possibility of involvement with one topic affecting message processing and
recall (see Appendix B). Mean anxiety scores for the animal (M = 38.5, SD = 13.42) and
rainforest group (M = 34.4, SD = 12.08) were subjected to an independent—samples t-test.
The results were non-significant, t (56) = 1.24, p = .221, n.s., indicating that no difference
in anxiety level existed between the two groups. In order to determine if messages were
perceived, in fact, as equally positive in nature, the means from the positive thought
listing for the animal (M = 3.64, SD = 2.3) and rainforest group (M = 3.14, SD = 2.34)
were subjected to an independent samples t-test. The results were non-significant, t (56)
= .812, p = .420, n.s., indicating that the two messages generated an equal number of
positive cognitions from subjects. Therefore message groups were collapsed for further
analyses.

Thought Listing

Participants were asked to list thoughts they had when reading the message. This
listing was not designed to prime participants into processing the message, but simply to
elicit the relative number of relevant thoughts that participants had after reading the
positive feeling message.

In this study, systematic processing is of concern. Systematic thoughts were
coded as those directly related to information contained in the message (Smith, Monison,
Kopfrnan, & Ford, 1994). Other thoughts related to message topic but not contained in
message information were coded as heuristic thoughts. Thoughts that were not related to
the message, such as “I am hungry” or “I feel tired” were coded as irrelevant. Two

coders coded all surveys and inter-coder reliabilities were calculated using Cohen’s

15

Kappa for systematic (K = .648), heuristic (K = .600) irrelevant (K = .783), and total
thought (K = .940) coding. All inconsistencies were resolved between the coders.
Memory Measure

The most popular, and possibly also the most simplistic, way to describe memory
is as a repository for information (Eysenck, 1977; Herrmann, 1996; Kleinmutz, 1966;
Koriat & Goldsmith, 1996; Loftus & Loftus, 1976; Morris & Gruneberg, 1994). Part of
information processing, memory encompasses the system of interaction between
cognitive processing and enviromnental stimuli involving storage and recall of
information. Memory is operationalized here as recognition of the positive feeling
message statements. Message recall was determined through scores on the message
recognition test (Appendix C).

All participants received a memory measure upon completion of the thought
listing. The memory measure contained a list of 10 randomly ordered statements _
regarding either the rainforest or animal website. Five of the statements had information
that appeared in the message, the other five contained information not presented
originally in the message. Participants were asked to assess whether each statement
appeared in the persuasive message they had just read. The instructions indicated that
some statements may not have appeared in the persuasive message. Thus, for
respondents in both conditions, five items could serve as “hits” (message statements
correctly identified as part of the original message) whereas the other five items could
serve as “false alarms” (new items incorrectly identified as part of the original message).

To firrther understand possible relationships in the data, Shapiro’s A’2 and B’3

statistics (Shapiro, 1994) were computed as measures of individual respondent sensitivity

16

and criterion bias, respectively. Briefly, A’ is used for calculating sensitivity, an
individual’s accuracy in recognizing something that they have seen before. While there
is no set range for A’, the larger A,’ the more sensitive the respondent is considered. B’
measures criterion bias, which can be considered a measure of a subject’s vigilance and
prioritization in identification tasks. B’ can range from +1 to -1, with a score of 0
indicating that the subject’s criterion is equally placed between the false alarm and hit
distribution. Higher scores indicate a subject’s greater concern for minimizing false
alarms at the cost of getting fewer hits. A score of —1 demonstrates the opposite — that
the subject was more concerned with maximizing hits than minimizing false alarms.
Results

Induction Check

The initial measures of state anxiety for both groups were submitted to an
independent samples t-test. The mean initial state score for the anxiety induction group
(M = 38.65, SD = 12.64) was not significantly greater than the mean state score for the
control group (M = 34.62, SD = 13.00), t (56) = -l .19, n.s., demonstrating no difference
in control and induction groups’ anxiety levels pre-induction, however the video camera
was present in the room for the anxiety induction group. Additionally, both the control
and induction groups’ initial state scale scores were submitted to single-sample t-tests
The two t-tests show that the mean initial state score M = 34.62) and the mean initial
induction score (M = 38.65) were not significantly greater than the normed mean for state
anxiety of 35.45, t (28) = -.34, n.s.; and t (28) = 1.37, n.s. respectively. This indicates
that neither the control group nor the induction group were more anxious than should be

expected initially.

17

In order to determine the effectiveness of the anxiety induction on state anxiety
levels the state STAI scores for the induction group pre and post-induction were
submitted to a matched pairs t-test. Respondents’ scores on the initial state scale (M =
38.65 , SD = 12.64) were not significantly greater than scores on the state scale after
completing the math test (M = 38.58, SD = 13.23, t (28) = .066, n.s.), indicating that the
induction failed to raise anxiety levels significantly. These results indicate that the
anxiety inductions of the math test did not significantly increase state anxiety in the
induction group.

To ascertain if post-induction scores were greater than the control groups’ state
scores, mean post-induction state scores (M = 38.5 8, SD = 13.23) and the mean control
groups’ initial state scores (M = 34.62, SD = 13.0) were subjected to an independent
samples t-test. Results from this test were non-significant, t (56) = -l.15, .255, n.s.,
however, group means were in the predicted direction. Since the expected difference
between groups was not significant, the established normed means for the state portion of
the STAI were used to split participants into high, moderate, and low levels of anxiety
according to their scores. Any score one standard deviation above the normed mean was
considered to be high (11 = 15), any score one standard deviation below the normed mean
was considered low (n = 16), and scores falling around the mean were considered
moderate (11 = 27). Thus, high low, and medium state anxiety distinctions were used for
further analysis.

ST AI
Normed mean scores exist for both state and trait anxiety and are as follows: for

state anxiety the normed mean is 35.45, SD = 10.5; for trait anxiety, the normed mean is

18

34.8 (SD = 9.22). The ranges for this study were within what would be expected from
previous literature.

Confinnatory Factor Analysis was nm on the state, trait, and full STAI. For the
full STAI, the two factor solution was found to be acceptable, with alpha reliabilities of
.926 for the trait factor and .958 for the state factor. Root Mean Square Error was
calculated for the two-factor solution and was an acceptable .0985. Internal consistency
was acceptable, the residual matrix did not have errors greater than expected. Chi-square
global test of fit was consistent with the two-factor solution, as it was statistically
significant when tested for a flat correlation matrix, x2 = 615.5, p < .00.

Cronbach Alpha reliabilities were within the parameters previously found for the
STAI and can be considered very good. Reliability for the initial state scale, anxiety
induction second state scale, and the final state scale ranged from or = .95—.96 .
Cronbach’s Alpha for the trait scale score was or = .93. As the state scale was used for
most analyses, Confinnatory Factor Analysis conducted on the state portion of the STAI
determined if all items were loading to one factor. Results from the CFA revealed that
indeed, all items on the state scale loaded to one factor, with an average in-cluster
correlation of .530 and a .958 standard score alpha.

Hypothesis One

Hypothesis one predicted that higher anxiety levels should correlate with an
increase in systematic processing. The Pearson correlation between anxiety and
systematic thoughts was found to be negative, but non-significant, r (58) = -.l76, p =
.187. A ratio of systematic to total thoughts for each respondent was then computed.

Results from Pearson correlations between state anxiety (using the control groups’ initial

19

scores and the induction groups’ post-induction scores) and the ratio of systematic to total
thoughts again displayed the opposite of hypothesis one’s prediction. A non-significant
correlation was also found between these variables, r (58) = -.209, p = .115, n.s..

AN OVA was used post hoc to assess possible differences in systematic
processing according to the levels of anxiety previously mentioned. Group means on the
dependent variables of thought ratio and systematic thoughts were contrasted. Contrast
procedures indicated that the high anxiety group was significantly lower in the number of
systematic thoughts produced than the moderate or low anxiety group, t (55) = 2.5, p <
.016. Additionally, the high anxiety group was significantly lower in their ratio of
systematic to total thoughts than the moderate or low group, t (55) = 2.86, p <.009.
Therefore the data does not support hypothesis one, and further, indicates a relationship
opposite to the predicted relationship between anxiety level and cognitive performance.
Means, standard deviations, and ranges for the three anxiety groups can be seen in Table
1.

Hypothesis Two

Hypothesis two postulated that as systematic thoughts increased, so too would
memory performance as measured by “hits.” The Pearson correlation coefficient
between hits and systematic thoughts was calculated and found to be both negative and
nonsigrrificant, r (5 8) = -.119, n.s. The A’ and B’ statistics calculated for all subjects
were correlated with systematic thoughts. The correlations between sensitivity and
systematic processing were non-significant (r (58) = .057, p = .628). Similarly, the
correlations between criterion bias and systematic thoughts were non-significant (r (5 8) =

-.015, p = .91). Thus, hypothesis two fails to receive support from these data.

20

nmv

 

 

 

8.78. 7 88. 78. 7 88. 78. 7 8. 88. K. we 8. e. 85 8:25
8.88. 8. 78. 8.88. 2. E. E. 3w. .8. S. C: asaasm
8.8 8.8 8.8 S. S. 3. E. 2. 2. 857 and
8. 78. 8.78. 88. 78.. 2. 8. mm. 88. 88. .8. ea
8.78 8.88 8.88 88. 83 Ba me. 82 :4 $889:
oflnfioumzm
8.7:. 8.78. 8.78. a. 8. fl. 8. mm. 88. 38. s “833
”888.8 582:.
0.50m
8-8 2.8 8.8 8.8 NS 2.8 8.88 2.3 8.3 base 23
885 08882 33 83m 2882 33 Em 28082 33 ssgomg

omqam commerce 582

85:58

 

833.8; to SKEMSQ BE .azohemEQ 3.538%. .38:

~ Each

21

Discussion

The Eflect of Anxiety on Systematic Processing

The results of this study fail to demonstrate that anxiety functions as a
motivational state to process positive feeling information as postulated by the NSRM.
High anxiety, working as a motivator, was predicted to increase systematic processing
and memory for positive feeling messages, as one cognitive symptom of anxiety is
narrowed attention. As the literature made a strong case for highly anxious individuals to
be on heightened alert regarding their environment, perhaps messages in a laboratory
setting are not the focus of their supposed narrowed attention. It may be that anxious
subjects would be better able to recall details of the laboratory itself and not the study.
This could be assessed in a future research. The possibility also exists that anxiety is not
a negative mood state, as originally thought. While previous research has not tested the
fimctions of anxiety in the HSM, positive mood has been tested extensively and results
show a heuristic processing bias. Therefore, assuming anxiety to be the opposite of a
positive mood, systematic processing was predicted to be the preferred pathway for
information. Whether anxiety can be accurately categorized as a negative mood state is
important to further understanding functioning within the HSM routes. Finally, it is
possible that subject’s ability to process systematically may have actually been impaired
by anxiety. In fact, as previously mentioned, Perez, Lopez & Woody (2001) detailed
symptoms of anxiety that implied impaired cognitive firnctioning, including difficulty
concentrating, narrowed attention, and distorted reasoning. As expected differences in
anxiety, memory, and processing were not found, post-hoe analysis was conducted to

determine if behavioral intention was influenced by subject’s memory performance.

22

Behavioral Intention and Memory Performance

The dual process models imply that attitudes formed through high elaboration
processes, such as those thought to occur in systematic processing are more predictive of
behavioral intentions (Petty, et.al., 1994; Slater, 2002). Therefore participant’s sensitivity
and criterion bias, both measures of memory fimctioning, were analyzed with the
behavioral intention measure to determine if a relationship existed.

Sensitivity (A’) and criterion bias (3’) were submitted to separate independent
samples t-tests, grouped by behavioral intention (see Table 2). Those who took the
behavioral intention measure, a website address on a bookmark that was stapled to the
experimental booklet, were significantly higher in both sensitivity (M = .88) and criterion
bias (M = .47) than those who did not take the behavioral measure (sensitivity and

criterion, respectively: M = .79, M = .067), t (55) = 2.32, p < .03; t (55) = 2.48, p < .017.

Table 2

Mean Table of Behavioral Intention, Criterion Bias, and Sensitivity

 

 

Mean Standard deviation Range
Behvaioral Intention Yes No Yes No Yes No
Sensitivity (A’) .88 .79 .09 .18 67-95 .25-1 .00
Criterion Bias (B’) .47 .07 .62 .62 -1.00-1.00 -1 .00-1.00

23

Further ad hoc analysis of the data revealed that there was a significant difference
in message type, animal or rainforest message (see Table 3 for means and standard
deviations), and sensitivity (A’ statistic), F (1) = 4.76, p < .035, but no significant
difference was found between message type and criterion bias (B’), F = (1) = .70, n.s.
That sensitivity and criterion bias were important in determining behavioral intention and
that significant differences existed in message type and sensitivity could imply that when
information is presented in terms of a persuasive message, involvement with topic may
assist in recollection of message information. In fact, Todorov (2002) in discussing the
three motivations to process information in the HSM, mentions defense motivation, an
important subset of involvement. Motivation to process influences processing route, and
perhaps in turn behavioral intention. If messages tap into areas of involvement, subjects
may process differently and recall more information. Again, this presents another
opportunity for further research to understand the relationship between memory measures

and behavioral intention in persuasive messaging.

Table 3

Mean Table of Message Type, Criterion Bias, and Sensitivity

 

 

Mean Standard deviation Range
Message Type Animal Rainforest Animal Rainforest Animal Rainforest
Sensitivity (A’) .80 .88 .18 .09 25-95 67-100
Criterion Bias (B’) .19 .33 .58 .73 -1.00-1.00 -1 .00-1 .00

24

In the dual process models, motivation is not the only determinant of processing
route. Ability also factors into a receiver’s choice of route. In this study, it was thought
anxious receivers would be motivated by their negative mood to relieve their mood state,
and process more systematically, according to the tenets of the NSRM. Since this did not
occur, it may be that processing ability of receivers who are anxious decreases. Future
research should vary message difficulty among anxiety levels to probe the motivation and
ability relationship in determination of message processing route.

Limitations

Finally, limitations of this study are discussed. In regard to the first hypothesis,
one possibility could be that relevant thoughts were not the outcome that should have
been measured with regard to systematic thoughts, and in fact hits should have possibly
been the ultimate outcome of interest. When using AN OVA to contrast the high anxiety
subjects (M = .75, SD = .17) with the low (M = .64, SD = .23) and moderate (M = .70,
SD = .21) on hits, it can be seen that high anxiety subjects do, in fact, have higher hit
rates, but they are not significantly higher than the other two groups.

In regard to participant ability and message difficulty, the messages were
analyzed post hoc to determine readability statistics. Microsoft Word (2000) was used to
calculate the Flesch Reading Ease (FRE) score and the F lesch-Kincaid Grade Level
(F KGL) score, two measures of readability. The FRE has a range of 0-100, with higher
numbers indicating easier text. An optimal F RE score for most audiences ranges from
60-70. The FKGL ranges from lit-12‘h grades with 7th or 8th grade levels considered best
for most audiences. The FRE scores for the animal and rainforest messages were 50.0

and 48.9, respectively. The FKGL scores for the animal and rainforest messages were

25

11.8 and 10.8, respectively. Both were outside of the range considered optimal for
general consumption, and could be considered more difficult reading. Thus, message
difficulty in this study serves as a limitation, as it could have hindered participant’s
ability to process messages. Future studies could address this issue.

Typical of much research conducted in the university setting, the sample size of
58 had a mean age of 21, all college students from the Midwest. The small sample size
was a limitation of this study. Additionally, in a study concerned with information
processing, using a college sample could possibly hinder ability to generalize from the
data, as college students are theoretically, at least, well practiced in focusing their
attention, and in the American education system, memorizing information for recall.
Indeed the mean A’ score (M = .835, SD = .14) for this sample was found to be
significantly different from 0, using a one sample t-test, t (57) = 42.7, p < .001. This
indicates that on the whole, subjects in this study were quite sensitive in their recognition
of something they have previously seen. Additionally, subjects were cautious on the
memory measure, as a one sample t-test using the mean B’ statistic (M = .26, SD = .64)
shows they were significantly concerned with minimizing false alarms, t (56) = 3.03, p <
.004. These results are to be expected from a group of subjects who are experienced test-
takers and familiar with performance assessment. Using a sample not as familiar with the
concentration required during assessment under high state anxiety conditions should, as is
the case here, hinder memory and systematic processing further. Furthermore, the
sample size should be larger to increase effect sizes and variance in the distribution of

many variables, although fairly equal group sizes were used for analysis.

26

To further understand the anxiety induction, measurement of other possible
variables, such as past performance on math exams and overall unease at completing
math tests should occur in future studies. In this study, there was no implication for
subjects on their math test performance and evaluation. Perhaps presenting the induction
task as having future implications for the participant would heighten anxiety and make
the induction more successful. Additionally, the induction should be longer and more
intense to heighten anxiety as much as possible. It may also be important to let more
time pass between the induction and second measurement of anxiety in order for the
induction to fully “sink in” with participants. In this induction, the camera was present in
the room from the beginning of the study, eliminating the possible anxiety-heightening
element of surprise. That participants were run in groups of five could have also reduced
anxiety levels (Schachter, 1959). In the future, inductions should be run individually.
Inducing anxiety is within itself, an area for future research.

Summation

In Stun, this study found that anxiety and systematic thoughts about a positive
feeling message were not significantly correlated as originally predicted, neither through
correlation of raw systematic thoughts nor through a ratio of systematic to total thoughts.
Additionally, contrasts of group means found that the high anxiety group was
significantly lower in the number of systematic thoughts produced than the moderate or
low anxiety group, t (44.49) = 2.5, p < .016), as well as in their ratio of systematic to total
thoughts than the moderate or low group (t (23.9) = 2.86, p < .009) which is opposite the
original hypothesized relationships. It was also found that systematic thought increase

was not related to memory performance as measured by “hits,” or by the A’ or B’

27

statistics. Correlations between systematic thoughts, hits, A,’ and B’ statistics were all
non-significant, showing that in these data, systematic processing was not related to

memory performance, contrary to what was predicted in hypothesis two.

28

APPENDIX A
COLLEGE MATH SKILLS TEST

1. If the radius of a wheel is f feet, how many revolutions does the wheel make per mile?
(1 mile= 5,280 feet)

a. 5,280f b. 2,640 c. 5,280nf
nf
(1. nt e._1tf

2,640 5,280

2. Base RT of triangle RST is 4/3 of altitude SV. If SV equals 6, which of the following
is an expression for the area of triangle RST?

a 2c b.2_c2 c 92

5 3 2
(1 4c2 6 8c_2
5 5

3. Four equal circles, each with a diameter of 1 ft. touch at four points as shown in the
figure below. What is the area in square feet of the white interior portion?

b. 1-1: c.1-41r d.1t e.

 

Al?!

4. A line segment is drawn from point (8, -2) to point (4, 6). The coordinates of the
midpoint of this line segment are:

a. (12, 4) b. (12, 8) 0. (6,4) (1. (6,2) e. (6, -2)

29

APPENDD( B
PERSUASIVE MESSAGES

DIRECTIONS:
Please read the following message. When you are done reading, you may move on to the
next page.

ANIMAL MESSAGE:

Over 10 million animals are put to death every year in the US. alone because they
are unwanted, abandoned, or abused. Many millions more are neglected or treated
cruelly. 27 million unwanted animals are given to shelters in the US every year. Last
year, visitors' clicks at The Animal Rescue Site funded 23,968,850 bowls of food for
animals in these shelters.

You can improve the lives of these animals for free -- the site's sponsors firnd the
purchase of a bowl of food to feed an animal awaiting adoption or living in an animal
sanctuary. You can feel good knowing you have contributed to helping this cause
through a simple click of your mouse. Depending on the speed of your modem, it might
take you only a few seconds each day to help provide care and food for an abandoned pet
or other animal by clicking the "Feed an Animal in Need" button. Each click on the
purple "Feed an Animal in Need" button at The Animal Rescue Site provides a bowl of
food for an animal at the world's largest pet adoption center, North Shore Animal League
America, or at one of the Fund for Animals’ world-renowned animal sanctuaries. The
Animal Rescue Site relies on its passionate supporters. Click every day to help and
encourage friends and family members to do so as well.

RAINFOREST MESSAGE :

Originally, 6 million square miles of tropical rainforest existed worldwide. As a
result of deforestation, only 2.6 million square miles remain today. The race is on to save
our rainforests and the incredible biodiversity they hold. A typical four square mile patch
of rainforest contains up to 750 species of trees, 125 mammal species, 400 species of
birds, and 150 types of butterflies. Last year, visitors' clicks at The Rainforest Site
funded a total of 246,852,687 square feet of rainforest.

You can help to save rainforest land for free -- the site's sponsors fund
preservation of rainforest land. You can feel good knowing you have contributed to
helping this cause through a simple click of your mouse. Depending on the speed of your
modem, it might take you only a few seconds each day to help preserve precious
rainforest land by clicking on the “Save Our Rainforests” button. Each click on the green
"Save Our Rainforests" button at The Save Our Rainforests Site helps preserve 11.4
square feet of rainforest land. Funds generated by your daily click go to the site’s land
trust partners: The Nature Conservancy and The Rainforest Conservation Fund. The
Rainforest Site relies on its passionate supporters. Click every day to help and encourage
friends and family members to do so as well.

30

APPENDIX C
MEMORY MEASURE

Below are statements regarding the message you just read about The Animal Site (The
Rainforest Site). Please indicate which statements were presented in the message you
read previously by placing an “X” on the line next to the appropriate statements. Some
statements were in the message and some were not. Only mark an “X” for the ones you
think were included in the message.
The Animal Site helps people adopt animals.
Ten million animals were put to death last year.
Clicking on the button is free.
The Site’s button is purple.
Each click funds a bowl of food.
Each click helps adopt an animal.
23 million bowls of food were given last year.
27 million bowls of food were given last year.

Each click firnds food at a local animal shelter.

23 million animals were saved last year.

RAINFOREST MEMORY MEASURE ITEMS:

__ The Rainforest Site helps companies sell land.
2.6 million acres of land exist today.
__ Clicking on the button is free.
__ The Site’s button is green.
__ Each click funds the preservation of land.
Each click helps care for rainforest animals.
11.4 square feet are saved with each click.
One acre is donated for each click.
Land is preserved by four government agencies.

The site’s button is red.

31

Footnotes

1In past research, whether a subject could distinguish weak from strong messages
was used as an indicator of which message-processing route occurred (Mackie and
Worth, 1989, 1991; Schwartz and Bless, 1991). Since the definition of “weak” and
“strong” messages is not adequately fleshed out and falls prey to tautological reasoning
(Mitchell, 2000, 2001), whether a subject can distinguish the two message types will not
be used as a primary form of measurement to determine if systematic or heuristic
processing occurred in this study.

2Formula for non-parametric measure of sensitivity (A’):

A'=1— .1. M + __|“P(")|

4 p0!) |1-p(FA)|

3 Formula for non-parametric measure of criterion bias (B’):
Br: P(hll-P(h)I-P(FA)I1-P(FA)l
P(h)|1-P(h)|+P(FA)ll-P(FA)I

 

32

References

Barnes, L.B.L., Harp, D. Jung, W.S. (2002). Reliability generalization of scores on the
Spielberger State-Trait Anxiety Inventory. Educational and Psychological
Measurement, 62 (4), 603-619.

Blanchette, I. & Richards, A. (2003). Anxiety and the interpretation of ambiguous
information: beyond the emotion —- congruent effect. Journal of Experimental
Psychology, 132 (2), 294-309.

Bless, H., Mackie, D. M., & Schwarz, N. (1992). Mood effects on attitude judgments:
Independent effects of mood before and after message elaboration. Journal of
Personality and Social Psychology, 63(4), 585-595.

Bless, H., Schwarz, N., & Strack, F. (1990). Mood and Persuasion: A cognitive response
analysis. Personality and Social Psychology Bulletin, 16(2), 331-345.

Bright, J EH & Freedman, O. (1998). Differences between implicit and explicit
acquisition of a complex motor skill under pressure: an examination of some
evidence. The British Journal of Psychology, 89 (2), 249-263.

Cattell, R. B., (1962). Advances in the measurement of neuroticsm and anxiety in
a conceptual framework of unitary-trait theory. Annals of the New York Academy
of Sciences, 93, 815-839.

Cattell, R.B., (1966). Anxiety and motivation: Theory and crucial experiments. In
C.D. Spielberger (Ed.), Anxiety and Behavior (23-59). New York: Academic
Press.

Eysenck, M.W., (1997). Anxiety and cognition. United Kingdom: Psychology
Press.

Freud, S. (1936). The problem of anxiety (H.A. Bunker, Trans). New York:
W.W. Norton. (Original work published 1936)

33

Greene, J. 0., Rucker, M. P., Zauss, E. S., & Harris, A. A. (1998). Communication
anxiety and the acquision of message-production skill. Communication
Education, 47(4), 337-347.

Goodwin, D.W., (1986). Anxiety. New York: Oxford University Press.

Hall, N. & Crisp, R. (2003). Anxiety induced response perseverance and stereotyping
change. Current Research in Social Psychology, 8 (17), 242-253.

Hamilton, M.A. & Hunter, J. E. (1988). Confirmatory Factor Analysis [Computer
software]. East Lansing: Authors.

Hodges, W. F., (1976). The psychophysiology of anxiety. In M. Zuckerman &
C.D. Spielberger (Eds), Emotions and anxiety: New concepts, methods and
applications (pp.175-194). New Jersey: Hillsdale.

Mackie, D. M., & Worth, L. T. (1989). Processing deficits and the mediation of positive
affect in persuasion. Journal of Personality and Social Psychology, 5 7(1), 27-40.

Mackie, D. M., & Worth, L. T. (1991). Feeling Good, But Not Thinking Straight: the
Impact of Postive Mood on Persuasion. In J. P. Forgas (Ed.), Emotion and Social
Judgments (Vol. 23, pp. 210-219). New York: Pergamon Press.

Microsoft Word 9.0 [Computer software]. (2000). Redmond, WA: Microsoft
Corporation.

Mitchell, M. (2000). Able but not motivated? The relative effects of happy and sad mood
on persuasive message processing. Communication Monographs, 67(2), 215-226.

Mitchell, M. M. (2001). The effects of anger, sadness and happiness on persuasive
message processing: A test of the negative state relief model. Communication
Monographs, 68(4), 347-359.

34

Mogg, K., Kentish,J., & Bradley, BR, (1993). Effects of anxiety and awareness on
colour identification latencies for emotional words. Behavioral Research and
Therapy, 31 (6), 559-569.

Nemiah, J.C., (1981). The psychoanalytic view of anxiety. In D.F. Klein & J .G.
Rabkin (Eds), Anxiety: New research and changing concepts (pp. 291-
300). New York: Raven Press.

Oei, T. P. S., Moylan, A., & Evans, L. (1991). Validity and clinical utility of the fear
questionnaire for anxiety-disorder patients. Psychological Assessment, 3(3), 391-
397.

Perez-Lopez, J. R., & Woody, S. R. (2001). Memory for facial expressions in social
phobia. Behaviour Research and Therapy, 39(8), 967-975.

Petty, R. E., Preister, J. R., & Wegener, D. T. (1994). Cognitive Processes in Attitude
Change. In J. Robert S. Wyer & T. K. Srull (Eds), Handbook of Social Cognition
(Vol. 2, pp. 69-142). Hillsdale, NJ: Lawrence Erlbaum Associates, Publishers.

Rimal, R.N., Morrison, D., Turner, M.M. (2003, May). The risk perception attitude
(RPA) framework 11: Testing the role of anxiety as an underlying mechanism.
Paper presented at ICA, San Diego.

Ruiz-Caballero, J. A., & Berrnudez, J. (1997). Anxiety and attention: is there an
attentional bias for positive emotional stimuli? The Journal of General
Psychology, 124(2), 194-210.

Russell, J. A., & Pratt, G. (1980). A description of the affective quality attributed to
environments. Journal of Personality and Social Psychology, 38(2), 311-322.

Schachter, S. (1959). The psychology of affiliation; experimental studies of the sources
of gregariousness. Stanford: Stanford University Press.

Scheier, I.H., (1962). Experimental results to date from the standpoint of the clinician.
Annals of the New York Academy of Sciences, 93, 840-856.

35

 

Schwarz, N., & Bless, H. (1991). Happy and Mindless, But Sad and Smart? The Impact
of Affective States on Analytic Reasoning. In J. P. Forgas (Ed), Emotion and
Social Judgments (V 01. 23, pp. 55-71). New York: Pergamon Press.

Shapiro, MA. (1994). Measuring recognition memory. In A. Lang (Ed.), Measuring
Psychological Responses to Media Messages (pp 133-148). Hillsdale, NJ:
Erlbaum.

Slater, M. D. (2002). Involvement as Goal-Directed Strategic Processing. In J. P. Dillard
& M. Pfau (Eds), The Persuasion Handbook: Developments in Theory and
Practice (pp. 175-194). Thousand Oaks: Sage.

Smith, S.W., Morrison, K., Kopfrnan, J .E., Ford, LA. (1994). The influence of prior
thought and intent on memorability and persuasiveness of organ donation
message strategies. Health Communication 6 (1), 1-20.

Spielberger, CD, (1966). Theory and research on anxiety. In C.D. Spielberger (Ed.),
Anxiety and Behavior (pp.3-19). New York: Academic Press.

Spielberger, CD. (1999). Measuring anxiety and anger with the State-Trait
Anxiety Inventory (STAI) and the State-Trait Anger Expression Inventory
(STAXI). In M. Maruish (Ed. ) The use of psychological testing for treatment
planning and outcome assessment (2"d ed.) (pp. 971 -.992) New Jersey: Lawrence
Erlbaum Associates.

Terry, W. S., & Burns, J. S. (2001). Anxiety and repression in attention and retention.
The Journal of General Psychology, 128(4), 422-432.

Todorov, A., Chaiken, S., & Henderson, M. D. (2002). The Heuristic-Systematic Model
of Social Information Processing. In J. P. Dillard & M. Pfau (Eds), The

Persuasion Handbook: Developments in Theory and Practice (pp. 195-211).
Thousand Oaks: Sage.

Weary, G., & Edwards, J. A. (1994). Social Cognition and Clinical Psychology: Anxiety,
Depression, and the Processing of Social Information. In J. Robert S. Wyer g& T.
K. Srull (Eds), Handbook of Social Cognition (V 01. 2, pp. 289-338). Hillsdale:
Lawrence Erlbaum Associates, Publishers.

36

Wells, A. (1999). A cognitive model of generalized anxiety disorder.
BehaviorModification, 23(4), 526-555.

37

  

  

   

l 5

(111111111)le

llllllljjllllljflll

3