Tit #98115:
LIBRARIES

MICHIGAN STATE UNIVERSITY
\ ‘ EAST LANSING, MICH 48824-1048 .
a L
Q! ‘ J )
109.; we I53

This is to certify that the
dissertation entitled

“Why should I trust you?” The Path from Information Valence
to Uncertainty Reduction, Cognitive Trust,
and Behavioral Trust.

presented by

Jina H. Yoo

has been accepted towards fulfillment
of the requirements for the

PhD degree In Communication

Major Professor’s Signature

0 Chm “11200“!

Date

 

MSU is an Afiinnative 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

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/CIRC/DateDue.p65-p.15

 

“WHY SHOULD I TRUST YOU?” THE PATH FROM INFORMATION VALENCE
TO UNCERTAINTY REDUCTION, COGNITIVE TRUST,
AND BEHAVIORAL TRUST
By

J ina H. Yoo

A DISSERTATION
Submitted to
Michigan State University

in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Communication

2004

ABSTRACT
“WHY SHOULD I TRUST YOU?” THE PATH FROM INFORMATION VALENCE
TO UNCERTAINTY REDUCTION, COGNITIVE TRUST,
AND BEHAVIORAL TRUST
By
Jina H. Yoo
This research examined how valence of information affects relational outcomes

in initial interaction. The relational outcomes of interest here are uncertainty reduction
and the states of both cognitive trust and behavioral trust. Uncertainty level and the
cognitive state of trust should act as the preconditions for behavioral trust, as behavioral
trust involves actual risk in a vulnerable situation rather than simply a perception of the
other individual. In order to examine the causal relationship between information
valence and multi-dimensional aspects of trust, an experiment was conducted.
Participants were randomly given either positively or negatively valenced information
about a confederate and were asked to fill out questionnaires assessing their uncertainty
level and cognitive state of trust toward the confederate. The Prisoner’s Dilemma game
was used to operationalize the construct of behavioral trust. The results indicated that
valence of information had a significant effect on the cognitive level of trust, especially,
perceived trustworthiness. However, information valence did not have a significant
effect on predicting behavioral trust, except for the positively valenced information
condition. The path model indicated that both uncertainty reduction level and perceived

trustworthiness were significant predictors for behavioral trust when the individual was

exposed to a piece of positive information about the partner.

Cepyright by
JINA H. YOO
2004

ACKNOWLEDGMENTS

Dr. Sandi W. Smith... the one who gave me my present and my future

Dr. Chuck Atkin, Dr. Tim Levine, Dr. Chuck Salmon Dr. Heesun Park. . .the ones who
gave me great knowledge and insights

Soony, Songsong, Kate, Songyi.. the ones who gave me great encouragement

J eesun, Jeesoo.. the ones who gave me laughter and comforts

J ung-Min.. the one who gave me a rest

New Hope Baptist Church... the place in which I found peace

Rachmaninoff, Beethoven, Chopin. the ones who gave me energy and inspiration

Dad, Morn, Jae, and Brian.. the ones who gave me love and always prayed for me '

And

God.. the one who gave me my life and my new life

I thank you all.
I dedicate this humble dissertation to you.

2004. 10.

7l-°- BUILHBlo-‘E Olh‘é'
-Pr§2]

iv

TABLE OF CONTENTS

LIST OF TABLES .............................................................................. vi
LIST OF FIGURES ............................................................................. vii
INTRODUCTION .............................................................................. 1
CHAPTER 1: LITERATURE REVIEW
Uncertainty Reduction Theory and Information Valence .......................... 2
Uncertainty Reduction and Trust ...................................................... 7
Cognitive State of Trust ................................................................ 9
Behavioral Trust ......................................................................... 12

CHAPTER 2: METHODS

Overview of Design ..................................................................... l6
Participants ................................................................................. 17
Stimulus Materials ....................................................................... 17
Procedure ...................................................................................... 20
Dependent Variables ..................................................................... 22
CHAPTER 3: RESULTS
Confirrnatory Factor Analysis ......................................................... 23
Induction Check ........................................................................... 24
Hypotheses ................................................................................... 26
Path Analysis ................................................................................ 28
Evaluation of the Alternative Models ................................................ 30

CHAPTER 4: DISCUSSION

Implications from Proposed Hypotheses ............................................. 31
Implications for Dimensionality of Negativity and Deviance ..................... 34
Recommendations for Scenario Study and its Ecological Validity ............... 35
Limitations and Future Research ...................................................... 37
TABLES & FIGURES ........................................................................... 4O
APPENDICES ........................................................................................ 45
BIBLIOGRAPHY .................................................................................. 5 1

LIST OF TABLES

Means and (Standard Deviations) for Each Dependent Variables Broken Down By
Information Valence Type .......................................................................... 40

Zero Order Correlations among Variables ........................................................ 41

Vi

LIST OF FIGURES

PD Game ........................................................................................... 42
Initial Path-Analytic Model ...................................................................... 43
Revised Path-Analytic Model .................................................................... 44

vii

“Why should I trust you?” The Path from Information Valence to
Uncertainty Reduction, Cognitive Trust, and Behavioral Trust.
Then you would trust, because there is hope, and you would look around and rest
securely. . .
KJV Job 11:18

In daily life, situations exist where, within a short time frame, people who are
strangers need to decide whether to trust others or not. People may not be aware how
often they have to make a trusting decision within initial interactions. Exchanges often
take place between strangers who cannot depend on past behavior or the prediction of
future interactions to build mutual trust. For instance, in the matters of hiring a babysitter
for a young baby, or purchasing a used-car from a stranger, people have to make
important decisions in their initial interactions. In these moments, when a trust decision
should be made swiftly within an initial interaction, how does uncertainty level affect
forming an impression about the person, making a judgment of trust, and acting in a
trusting manner?

Studies on uncertainty reduction have tested the relationship between uncertainty
reduction and social attraction as a positive relational outcome (Douglas, 1990; Harvey,
Wells, & Alvarez, 1978; Planalp & Honeycutt, 1985). However, little attention has been
given to the relationship between uncertainty reduction and trust, except for a few studies
indicating uncertainty reduction and trust in romantic relationships (Clatterbuck, 1979;
Gudykunst, 1985; Planalp & Honeycutt, 1985; VanLear & Trujillo, 1986). These studies,
nevertheless, only focus on the cognitive state of trust toward the other partner, mostly
assessed through self-report questionnaires. Actual behavioral trust was not measured in

these studies. Whereas many scholars in social science argue that the construct of trust

consists of cognitive, emotional, and behavioral dimensions (Barber, 1983; Bern, 1970;
Deutsch, 1958), few studies in the communication field have tested the behavioral aspects
of trust. Therefore, the present research will investigate the relationship between the
cognitive state of trust and actual behavioral trust in initial interaction.

This study will focus how valence of information may affect uncertainty
orientation and the cognitive state of trust for a target person, and how this will affect
willingness to risk by engaging in trusting behavior. Specifically, the investigation seeks
to explicate the relationship among information valence, uncertainty reduction, the
cognitive state of trust, and behavioral trust through a path model specifying the
relationship among these variables.

Uncertainty Reduction Theory and Information Valence

The original URT by Berger and Calabrese focused exclusively on the potential
influence of uncertainty and reduction of uncertainty during the beginning of a
relationship (Sunnafrank, 1986). URT focuses on the initial phase of interaction between
strangers. It states that when strangers meet, the primary motivation is to reduce
uncertainty and increase predictability about the behaviors of themselves and the others
(Berger & Calabrese, 1975). When individuals engage in initial interactions with other
individuals, a certain degree of uncertainty always exists. Through communication,
individuals exchange information about each other, and information usually leads to
decreases in uncertainty. According to Berger’s definition, uncertainty about the other is
the “(in)ability to predict and explain (his or her) actions” (1987, p.41). Uncertainty can
be also Viewed as a “phenomenon” that is relevant to interpersonal communication,

which explains communicative behavior in terms of an individual’s inability to

understand both his/her own and the other person’s attitudes, feelings, and behaviors
(Berger and Calabrese, 1975). Berger’s work on uncertainty specifically states:
The problem in initial interaction is one of predicting the likely behaviors of the
other person or persons in the situations so that one can select from his
behavioral repertoire those alternative modes of behavior which are deemed most
appropriate to the other person or persons in the Situation. . . Uncertainty is
reduced by generating and confirming predictions and constructing and verifying
explanations for the behavior of one’s self and the other(s) (1979, p.123-124).
Thus, the basic premise of URT is that when strangers encounter each other for the first
time, they try to reduce uncertainty and to increase predictability about each other’s
behavior. In other words, individuals are motivated to reduce uncertainty, since they find
the state of uncertainty uncomfortable, and as a result, reducing uncertainty is expected to
enhance individuals’ perceived ability to forecast future relational outcomes.

Holmes and Rempel (1989) indicated that the state of uncertainty increases
interference with normal relational development. By attempting to reduce uncertainty, an
individual can increase predictive certainty about the other’s future behavior as well as a
certainty about the self in the situation. The ability to verify the other’s behavior
alleviates anxiety and vulnerability that comes from high uncertainty (Gudykunst, 1985).
This line of reasoning leads to the argument that the reduction of uncertainty is associated
with positive outcomes in the initial interaction, such as liking (Douglas, 1990), by
assuaging the negative affect caused by uncertainty toward both the other and to self. If
individuals are unable to “get to know” each other, the possibility they will develop a

more enduring relationship is reduced (Berger & Calabrese, 1975). Failure to reach a

certain point of uncertainty reduction means that a relationship cannot develop beyond
the most superficial levels (Berger & Gudykunst, 1991). In general, high uncertainty
about acquaintanceship is associated with relatively negative views on initial interaction,
increased avoidance of initial conversation sequences, and relative inability to manage
acquaintanceship episodes successfully (Douglas, 1991).

In general, a line of research on URT suggests that positive relational outcomes
result from uncertainty reduction about another individual, while negative outcomes
come from high states of uncertainty. Gudykunst and Kim (1997) argued that individuals
have maximum and minimum thresholds of uncertainty, so that if individuals’ unceItainty
levels exceeds their maximum thresholds, they may not have enough information to
communicate, and cannot predict what the other’s behavior would be, therefore they feel
uncomfortable communicating. In such cases, individuals lack confidence in their ability
to predict the other’s behavior, thus, feeling vulnerable, these individuals have difficulties
processing information. Similar findings have suggested that high uncertainty leads to
negative outcomes in the relationship, such as relational disintegration (Harvey et al.,
1978), negative relational consequences, especially in a romantic relationship (Planalp &
Honeycutt, 1985), and decreases in attraction (Douglas, 1990).

However, reducing uncertainty does not always guarantee positive relational
outcomes. Although the original axioms state that uncertainty reduction is related to
positive relational outcomes, Berger (1986) later argued that individuals may actually
learn that their interactional partners possess attributes that make them less attractive
while reducing uncertainty; that is the point at which one attempts to stop reducing or

even to increase uncertainty. Similarly, Schcidel (1977) claimed that, in the process of

reducing uncertainty, individuals might discover undesirable facts about their partners,
then uncertainty reduction actually brings negative relational outcomes. Kellermann and
Reynolds (1990) also indicated when an individual perceives another as satisfying certain
needs, the prediction will confirm that he/she is more likely to perceive another as more
positive and rewarding, therefore more uncertainty reduction as well as positive relational
outcomes will occur. On the other hand, when the individuals’ behaviors or attributes are
deviant, these violate expectations and social norms, then uncertainty toward this person
increases in spite of having increased amount of communication; as a result, negative
relational outcomes will be established. I

These findings suggest that uncertainty reduction up to a certain level is a ‘
necessary, but not a sufficient, condition for achieving positive relational outcome. The
most important key to reduce uncertainty and maximize the positive relational outcomes
is the valence of the information received. In other words, positive confirmation of the
other individual while getting to know him or her would be obligatory to ensure further
uncertainty reduction as well as positive relational outcomes. By contrast, when
individuals’ behaviors or attributes are perceived to be negative, it produces negative
relational outcomes, such as an attempt to restrict or terminate an initial interaction
(Sunnafrank, 1990).

During the initial interaction, a piece of negative information about an individual
tends to be weighted more heavily than comparable positive information, which is called
the “negativity effect” (Kellermann, 1989). Numerous studies have found that even if the
number of positive and negative attributes is controlled and balanced, negative

impressions tend to emerge (Anderson, 1965; Brinbaum, 1973; Fiske, 1980; Oden &

Anderson, 1971; Reeder & Coovert, 1986; Van der Pligt & Eiser, 1980). For instance,
when participants were presented with equal amounts of polarized positive and negative
information about a hypothetical stranger, the negative information had significantly
more influence on character rating than the positive information (Richey, McClelland, &
Shimkunas, 1967). Richey and colleagues (1975) extended their research by varying the
amounts of favorable and unfavorable information used to describe behaviors of an
unknown person. Researchers increased the amount of positive information, and found
that a single negative behavior neutralized five positive behaviors. This suggests that
negative information is given more weight in information integration, thus yielding a
negativity effect.

Kellermann and Reynolds (1990) conducted a series of experiments on
uncertainty-provoking situations, and found that when the target’s behavior became more
deviant, level of uncertainty increased. As a result, in terms of relational outcomes, their
findings suggested that “the more atypically someone behaves. . . the less we want to talk
to them” (1990, p.66). A similar line of reasoning applied to this study predicts that
finding out a piece of negative information while forming the initial impression will be
viewed as highly deviant and socially undesirable. It will increase the uncertainty level.
Following this reasoning, the first hypothesis is forwarded:

H1: In initial interaction, participants exposed to positive information about the
partner will exhibit the highest uncertainty reduction level, followed by participants
exposed to no information, and participants exposed to negative information.

When an individual reaches a certain level of predictability regarding the future

behavior of other individuals via the uncertainty reduction process, and if the

predictability and expectation are positively confirmed by the others’ behaviors or
attributes in initial interaction, then a positive relational outcome, such as trust, is likely
to be established. Therefore, the next step is to examine the relationship between
uncertainty reduction and both the cognitive state of trust and behavioral trust.
Uncertainty Reduction and Trust

In general, trust is defined as a psychological state comprising the intention to
accept vulnerability based upon positive expectations of the intentions or behavior of
another (Rousseau, Sitkin, Burt, & Camerer, 1998). This notion is in accordance with a
definition of relational trust by Hosmer (1995), that he defined it as an individual’s
optimistic expectation of other’s behavior, when the individual must make a decision
about how to act accordingly. Trust also refers to the notion of the degree one risks of
himself/herself, concerning the other individual: this risk is predicated on the belief that
the partner is beneficent and dependable (J ohnson-George & Swap, 1983). Mayer et a1.
(1995) used the notion of expectation while defining trust, i.e. that trust is the willingness
of a trustee (person on whom trust decision focuses) to be vulnerable to the actions of
another party based on the expectations that the other party will perform a particular
action important to the truster (a person making trust decision).

When monitoring these various conceptual definitions of trust, it should be noted
that three basic conditions of trust emerge: (1) uncertainty, (2) vulnerability, and (3)
predictability. These three constructs are utilized to understand the complex process of
trust and uncertainty reduction. Pearce (1974) Specifically explained that there should be
some form of “contingency” between the truster and the trustee. His notion of

“contingency” is the same notion as a certain degree of “uncertainty” toward the other

individual in the communication process. If no uncertainty exists between two
individuals, it indicates that no risk or threat is found in future interaction with this
individual (Miller & Rogers, 1987; Pearce, 1974). Since it is virtually impossible to have
absolute uncertainty-free interaction (in other words, a certain degree of uncertainty
always exists in social interactions), one needs to make an effort to reduce uncertainty
and to increase predictability about how the other individual will act while
communicating with others. If an individual perceives high uncertainty toward the other
interactant, he/she comes to feel vulnerable, which will result in a negative relational
outcome. A vulnerable position is generated where one’s desired outcome is less
rewarding to his or her ego than the potential cost of an action (Miller & Rogers, 1987).
When perceived vulnerability is high, no basis for the development of trust will be
established. To prevent generating a vulnerable position, the truster should have some
confidence in the prediction of the other’s behavior. The predictability concerns the
individual’s attempt to seek and develop dependable sources of reward (Miller & Rogers,
1987). Miller and Rogers (1987) suggested that predictability functions as producing a
sense of certainty about future actions of the other individual. Without a certain degree
of predictability, an individual has no basic assumption of how the other individual will
or will not utilize his/her trusting behaviors (Pierce, 1974). When attempting to create a
sense of certainty about future action, the previous step would be the process of
“uncertainty reduction,” implying that as uncertainty (i.e. contingency) toward an
individual is reduced, one’s predictability of the other individual’s future action increases
in turn, and this eventually decreases one’s perceived vulnerability.

Therefore, uncertainty reduction is a necessary condition for the development of

relational trust. In order to make uncertainty reduction a sufficient precondition for
development of relational trust, exposing to a positive information should be the first step
in uncertainty reduction. One’s predictability about the other should be increased thus
reducing uncertainty via communicating with the other. As a result, when more
uncertainty is reduced, perceived predictability should be increased, vulnerability will be
minimized, and positive relational outcomes will be likely to develop. Two possible
relational outcomes are cognitive and behavioral trust.

According to previous research, interpersonal trust can be conceptualized and
operationalized in two different ways: (1) favorable perceptions of others in terms of
credibility and character (Berlo, Lemert, & Mertz, 1969; Giffin, 1967; McCrosky, 1966)
and (2) dependent or risky behaviors which indicate trust (Pearce, 1974). The first
conceptualization is called the “cognitive state of trust” while the second is called
“behavioral trust” (Pearce, 1974). Wheeless and Grotz (1977) explicated these two
concepts and state that interpersonal trust should be more completely conceptualized as a
process of engaging in certain types of dependent behaviors related to favorable
perceptions of the trustworthiness of another individual in a risky situation, where the
expected outcomes dependent upon that other person are not known with certainty.
Therefore, trust should be measured as both levels of perceptions and behaviors.
Cognitive State of Trust

Pearce (1974) differentiated the cognitive state of trust from trusting behavior by
specifying that a cognitive state of trust exists in one’s perception, so one individual
“assumes” or infers that the other’s behavior will not confer unacceptably negative

outcomes upon him or her. This assumption largely depends on the individual’s

perception of the other’s knowledge, character and motives. Although the dimensionality
of cognitive state of trust construct is still a controversial issue, generally three
dimensions are found in the construct of ethos/source credibility: (1) competence, (2)
goodwill, and (3) trustworthiness.

99 6‘

Competence, sometimes called “expertise, expertness”, or “authoritativeness”,
is the extent to which an individual is in a position to know the truth, to know what is
right or correct (O'Keefe, 1990). The second dimension in credibility is “goodwill” or
“intent toward receiver” (McCroskey & Teven, 1999, p.90). Ifone perceives “goodwill”
in another individual, the perceiver is willing to risk him or herself because the trusted
individual is perceived as benevolent (Canary & Cupach, 1988). Pearce (1974) pointed
that perceived “well-intention” means that the trustee is perceived as motivated to
conduct himself/herself such a way as to avoid negative outcomes for the truster.
Benevolence (Lazelere & Huston, 1980), altruism (Sheppard & Sherman, 1998), and
caring (McCroskey & Teven, 1999) are desirable qualities of perceived goodwill.
McCroskey and Teven (1999) described goodwill as “a means of opening communication
channels more widely (p.92)” that people would listen more attentively to a person whom
they believe has their best interests at heart.

Perceived trustworthiness is a belief by a person in the integrity of another
individual (Larzelere & Huston, 1980). This dimension in cognitive state of trust is the
assessment of whether the individual will likely be inclined to tell the truth as he/she sees
it (O’Keefe, 1990). Also a partner would be seen as trustworthy if he/she is the type of

person who can be counted on to be honest and reliable (J ohnson-George & Swap, 1983;

Larzelere & Huston, 1980; Rempel, Ross, & Holmes, 2001). Perceived trustworthiness

10

requires that one’s partner have the attributes that alleviate against cheating, abusing, or
harming the self-esteem of others (Sheppard & Sherman, 1998). In organizations,
trustworthiness is more central to the willingness to defer to authorities than is
competence (Lewicki & Bunker, 1995). Honesty and integrity are regarded as
components of perceived trustworthiness, since they are attributes of partners unlikely to
take advantage of opportunities to cheat (Butler, 1994; Larzelere & Huston, 1980).

Determining the attribution concerning a partner’s benevolence and integrity is
important for assessing the future potential of a relationship (Altman & Taylor, 1973;
Deutsch, 1958; Giffin, 1967; Larzelere & Huston, 1980). In other words, the more
individuals perceive benevolence and integrity in the target person, the more likely those
individuals will be able to predict a favorable future outcome for the relationship
(Larzelere & Huston, 1980). Applying URT to the cognitive state of trust, if a piece of
information given during the initial interaction positively confirms the individual, the
uncertainty level will further decrease toward the interactant, as a result, it will increase
positive relational outcomes, such as cognitive state of trust (Gudykunst, 1985; Planalp
& Honeycutt, 1985).

Hamilton and Zanna (1972) predicted that in forming first impressions of a
stranger, of a primary concern is to determine whether or not he/she will pose a potential
threat to future interaction. If a piece of information about the other individual is
negatively confirmed, an individual will be likely to stop reducing uncertainty toward an
interactant, and perceives this person would pose a potential threat to future interaction,
the cognitive state of trust will not be developed.

Therefore, when behaviors or attributes of others positively confirm an

11

individual’s prediction of the other while still reducing uncertainty toward him/her, the
cognitive state of trust toward the other will be established as a consequence of predicting
a positive relational outcome. When negative confirmation is perceived, most individuals
are not likely to engage in developing the cognitive state of trust, as a result, individuals
will stop reducing, or further increase, uncertainty levels. Therefore, the following
hypothesis, regarding the relationship between cognitive state of trust and uncertainty
reduction level, is presented:

H2: In initial interaction, participants exposed to positive information about the
partner will have the highest scores on the cognitive state of trust, followed by
participants exposed to no information, and participants exposed to negative information.
Behavioral Trust

One’S trusting behaviors are active since they involve putting oneself voluntarily
in a vulnerable situation in which the other actually might cause good or harm to him/her
(Pearce, 1974). In perceived trust, one’s vulnerability is minimal and limited, but trusting
behavior involves the voluntary action of making oneself vulnerable to the decisions of
others (Pearce, 1974). Therefore, behavioral trust requires an active process either of
putting oneself in a risky position or of escaping from putting oneself in such a position.
An individual can win if he/she trusts, but he/she can also lose by such trust. The loss
caused by trusting others varies broadly, in that it might be trivial, such as losing an
insignificant amount of money. However, when the loss directly interferes with one’s
own benefits or welfare, it might be disturbing and the costs are high. Therefore, trusting
behaviors always involve the components of uncertainty and vulnerability, and they

imply that one might lose by trusting the other.

12

AS a method for studying behavioral trust, the Prisoner’s Dilemma (PD) game is
often utilized. The Prisoner’s Dilemma (Rapoport, 1965) is a two-person non-zero-sum
game, in which the gains or losses incurred by each individual are a function of the
choices made by one’s partner as well as the choices made by one’s self (Deutsch, 1958).
In this game, two individuals must make dichotomous choices, the totality of choices
determines the outcome of the game, and it is assumed that the rank order of preferences
for the outcomes is different for different players. The typical scenario for this game is
described as following:

Imagine that you and an accomplice have been charged with having committed a

crime and are now prisoners, sitting in separate cells, unable to communicate,

and awaiting trial. The prosecutor offers a deal to persuade you to confess:

“There is enough circumstantial evidence to convict both of you, so even if you

both remain silent you both will be convicted and locked up for a year. But if

you admit your guilt and help to convict your silent partner, you will go free and
he will be put away for ten years. The reverse happens, of course, if he confesses
and you remain silent; and if you both confess, then unfortunately you both will

get nine years” (Richmond, 1985 , p.4).

The game is illustrated in Figure 1. Person I has to choose between rows X and
Y; Person II has to choose between columns A and B. Loss/gain will be determined by
the box that the other partner chooses interdependently. If Person I chooses column A
and Person II chooses row X, and both get into the AX box, they both have to live only a
year in prison. When examining the possibilities of choice for Person I, he/she can win

most and lose least by choosing row Y, confessing. Similarly, Person II can win most

13

and lose least by choosing column B. However, when both choose according to their best
interests, both have to live nine years in prison. Both have to be in the AX box to win
together.

The interests of the players are generally in conflict. The dilemma comes from
the fact that if both individuals choose the high-risk option (both remain silent), both
individuals receive positive outcomes, which is the most idealized situation; if both
choose the low-risk option (both confess), both receive negative outcomes; but if one
chooses the hi gh-risk option and the other chooses the low, the former receives a
maximized positive outcome and the latter a maximized negative outcome.

Deutsch (1958) also utilized PD as a method for studying behavioral trust; He
defined the three different situations as, cooperative, individualistic, competitive. In the
cooperative orientation, both participants feel that they are perceived as benevolent and
that they could place themselves in vulnerable positions. Under this orientation,
participants are likely to establish or perceive a relationship of mutual trust, thus leading
to behavioral trust. In the individualistic orientation, each participant feels that his only
interest was in doing as well for himself as he could, without regard to how well the other
person did. Under an individualistic orientation, participants are unlikely to establish or
perceive mutual trust since they have to make their choices. In a competitive orientation,
each subject was led to feel that he/she wanted to do better than the other, no matter what,
and each assumes the other would do the same. Under this orientation, participants will
be least likely to establish mutual trust in the relationship. Therefore, behavioral trust
tends to emerge only in a cooperative orientation in the PD game (Deutsch, 195 8).

The development of positive uncertainty reduction should be the basis for

14

engaging in cooperative orientation as well. When a piece of information about the
individual presented in initial interaction is positively confirmed, the uncertainty will be
reduced more, as a result, the chance of engaging in cooperative orientation in the game
will be increased. In contrast, where higher uncertainty levels exist between two
interactants, or a piece of information in initial interaction is negatively confirmed the
prediction, less perceived trust will exist, and the competitive orientation will be more
likely to be engaged.

In initial interaction, where most of the individuals lack prior information or
knowledge of the other individual, perceived “well-intention” and “benevolence” of
others via communication would be the only source they can rely on. Although Pearce
(1974) suggested that behavioral trust may or may not be accompanied by a cognitive
state of trust, this research will present a slightly different perspective that the cognitive
state of trust will have an impact on behavioral trust in initial relationships. When an
individual initiates communication with others for the first time, not enough background
information about strangers is present. Thus, the initial information given would be the
only information available for an individual to determine the cognitive state of trust of the
other. Cognitive trust built during the initial interaction would then function as the
primary source for individuals to voluntarily choose whether or not to be in a vulnerable
situation. Also, when an individual perceives negative relational outcome from the other
interactant, behavioral trust will not be established. Therefore, the following hypothesis
will be proposed:

H3: In initial interaction, participants exposed to positive information about the

partner will exhibit significantly more trusting behaviors, followed by participants

15

exposed to no information, and participants exposed to negative information.

Taking all these variables into consideration, a path model will be proposed that
will specify the relationship among uncertainty reduction, and two types of trust for the
positive information condition. In the positively valenced information condition,
uncertainly reduction is predicted to lead to the cognitive state of trust, which ultimately
leads to behavior trust. Negative information is assumed to lead to less uncertainty
reduction, and this state should not be related to cognitive or behavioral trust. The
proposed hypothesis is the following.

H4: A path model specifying the causal relationships among uncertainty
reduction level, cognitive state of trust, and behavioral trust, for the positively valenced
information condition will fit the observed data (see Figure 2).

Method
Overview of Design

This study examines the effect of information Valence on uncertainty reduction
and trust. In order to test the hypotheses proposed, a laboratory experiment design was
employed. The independent variable was the valence of information, which was actively
manipulated by trained confederates. Dependent (or mediating variables in terms of path
analysis) variables included (1) uncertainty reduction about the partner, which was
assessed using Clatterbuck’s (1979) CLUE7 scale (Appendix A), and (2) three
dimensions of cognitive state of trust, perceived competence, perceived goodwill, and
perceived trustworthiness predicted to be second order unidimensional, which was
assessed using McCroskey’s ethos/credibility scale (Appendix C). Behavioral trust was

the dependent variable, which was assessed via scores from the PD game (Appendix D).

16

To control for extraneous variables such as individual differences, (1) a tolerance for
uncertainly scale (Appendix B), and (2) a general communication suspicion scale
(Appendix E), was be measured by self-report questionnaires.
Participants

Ninety five students enrolled in an entry-level communication course at a large
mid-westem university earned course credits for participating in this study. About thirty
students were be randomly assigned to each of three conditions. No specific
characteristics were required for participation. All participants received course research
credit in exchange for their participation, and participation was voluntary.
Stimulus Materials

In order to ensure the quality of the study, a pretest was drawn from a sample of
185 respondents from the same population as the main study. The pretest consisted of
185 students who were enrolled in an undergraduate communication course at a large
Midwestern university, for which they received a research participation credit. The
sample was 67% female and had a mean of age of 20.2 (SD = 1.58). The participants
were asked to fill out an anonymous self-report survey, which typically took 20 to 30
minutes. The first section of the survey contained demographic questions, and the second
section contained self-reported attitudinal measures of the hypothetical person in a
hypothetically given situation.

The participants were randomly given one of two situations whereby they
interacted with either a potential blind date partner or a group project partner. Analyses
revealed no Significant difference between these two hypothetical persons (a blind date

partner and a group project partner) at an alpha level of .05, thus these two topics were

17

combined together for purposes of further analysis.

Five hypothetical situations were suggested from extensive literature reviews
(Brinbaum, 1973; Fiske, 1980; Oden & Anderson, 1971; Reeder & Coovert, 1986;
Richey et al., 1967; Richey et al., 1975; Van der Pligt & Eiser, 1980). They portrayed
whether an individual (1) had to attend Big Brother Big Sister meeting or had to attend
the court because of DUI problem, (2) helped a friend in need or did not help a friend in
need, (3) gave an extra change back to the restaurant or kept the change, (4) listened
carefully to other’s opinion or did not listen carefully to other’s opinion, and (5) found a
wallet on the street and reported it to police or kept it. Theses scenarios were used to test
the dependent variables in respect to: (l) perceived negativity, and (2) perceived deviance
of the given information. Each scenario had two conditions, a positively valenced
information condition and a negatively valenced information condition. In the pilot test,
respondents were asked to fill out scales after reading randomly selected hypothetical
scenarios. Participants in the negatively valenced information condition from the
“restaurant scenario” reported significantly higher scores on the perceived negativity
scale (M = 5.17, SD = 1.35) in respect to the control condition (M = 2.81, SD = .99, t (41)
= 11.34, p < .001). Participants who read positively valenced information reported
Significantly lower scores on the perceived negativity scale (M = 1.83, SD = 1.01) than on
the control condition, t (42) = - 6.37, p < .001. Perceived deviance for the positively
valenced information (M = 2.49, SD = 1.08) was not significantly different from the
control (M = 2.33, SD = 1.13). However, the negatively valenced information condition
reported a significantly higher score on the perceived deviance scale (M = 4.64, SD =

1.13) from the control condition, t (41) = 13.35,p < .001.

18

One-way AN OVA was conducted in order to determine whether these two
information valence conditions in the “restaurant condition” produced the intended
effects. The results Showed both a significant main effect for perceived negativity,
F(1,83) = 167.74, p < .01, 772 = .67, and a significant main effect for perceived deviance,
F (1,83) = 81.18, p < .01, 772 = .49. Therefore, the results from the pilot test indicated that
“restaurant condition” successfully induced the intended effects in respondents.

Participants in the negatively valenced information condition from the “wallet
scenario” reported significantly higher scores on the perceived negativity scale (M = 5.33,
SD = .46) in respect to the control condition (M= 2.81, SD = .99, t (5) = 5.54,p < .001).
Participants in the positively valenced information condition fi'om the “wallet scenario”
reported lower scores on the perceived negativity scale (M = 2.64, SD = 1.78) in respect
to the control condition, but no statistical significance was found. In terms of perceived
deviance, participants who read positively valenced information (M = 3.87, SD = 1.63)
were not significantly different from the control condition (M = 2.33, SD = 1.13).
However, negatively valenced information reported a significantly higher score on the
perceived deviance scale (M = 4.33, SD = 1.10) than the control condition, I (5) = 4.49, p
< .001.

One-way AN OVA was conducted in order to determine whether these two
information valence conditions in the “wallet condition” produced the intended effects.
The result showed a Significant main effect for perceived negativity, F (1,9) = 9.43, p
< .05, 772 = .51, however, no statistical Significance was found for perceived deviance.
Therefore, the pilot test indicated that “wallet condition” partially induced the intended

effects in respondents.

19

Procedure

Participants were required to Sign up with a person (confederate) whom they had
never met before. Upon arrival at the laboratory, the confederate pretended that he/she
was one of the participants. When a participant and a confederate entered the laboratory
room, the experimenter greeted them, provided an overview of the procedures, and
secured informed consent. Before each session started, the experimenter provided a brief
overview about the experiment, namely that this experiment was designed to see how to
build trust between two strangers. Then the participant was asked to fill out the general
tolerance for the uncertainty scale (see Appendix B), and the general suspicion scale (see
Appendix E). Upon of these questionnaires, the experimenter brought a confederate into
the experimental room, and gave a participant and a confederate a different question list.
During the interaction, a participant and a confederate could only answer the questions
that were asked by their partner. The question list, which mostly consisted of
demographic information, was carefully scripted, whereby the confederate was trained to
answer exactly the same way in each interaction.

In treatment conditions using the “restaurant condition”, the participant asked a
confederate: “Suppose you went to a restaurant, and the cashier mistakenly gave you an
extra $10 in change, what would you do?” In the positively valenced information
condition, the confederate answered: “Well, that happened to me once, and I gave it
back.” In a negatively valenced information condition, the confederate would answer:
“Well, one time when it happened I kept it.” In treatment conditions using the “wallet
condition”, the participant asked a confederate: “Suppose you picked up the wallet on the

street, and there were some cash and possibly credit cards, what would you do?” In the

20

positively valenced condition, the confederate answered: “Well, that happened to me
once, and I reported it to police.” In a negatively valenced information condition, the
confederate would answer: “Well, that happened to me once, and I ended up keeping the
wallet.” In a control condition, this step was skipped and the participant did not receive
any further information about the partner (confederate). Each dyad, which consisted of a
participant and a confederate, was randomly assigned to one of three information
valenced conditions.

After the scripted interaction, the induction check (see Appendix F) was
conducted to see whether the given information during the interaction induced intended
effects. Then, the participant responded to self-report questionnaires that included a
measure of specific uncertainty reduction about their partner (Appendix A), and three
different dimensions of the cognitive state of trust, which consist of perceived
competence, perceived goodwill, and perceived trustworthiness (Appendix C). In order
to prevent an order effect, multiple versions with various orders were employed.

Afterward, the experimenter described the PD game and its score distributions.
Before starting the games, the experimenter emphasized that this was a real situation
whereby their gain/loss would be directly related to a SS 0.00 value of gift certificate. In
other word, if they won the game, they would receive a chance to win a $50.00 gift
certificate, if they lost the game, they would not have a chance to receive the $50.00 gift
certificate.

Three-trial games were conducted. The format of the game was exactly the same
across three trials, only the point distributions differed. As the level of game increased,

the point distributions became more extreme. Each time, the experimenter asked both the

21

confederate and the participant for their decision to be written on paper. After
completing three PD trials, the experimenter debriefed those participants who were
involved in the experiment.

Dependent Variables

The uncertainty level (Appendix A) was measured by using a slight variation of
CLUE7 (Clatterbuck, 1979, cited in Kellermann & Reynolds, 1990). The variation was
made by Kellermann and Reynolds (1990), in that the eighth item (i.e., How well do you
think you understand the person?) was included. The Specific uncertainty scale measured
interaction-specific uncertainty which specifically asked uncertainty orientation about the
interactant on a series of 7-point bipolar scales. Clatterbuck’s CLUE7 scale has been
employed in nearly all uncertainty research, since it is a fairly reliable measure
throughout various studies (alpha ranged from .87 to .91).

Tolerance of uncertainty scale was measured in order to control for individual
differences on uncertainty orientation. The scale was originally designed by Clatterbuck
(1979), and revised by Kellermann and Reynolds (1990) to measure one’s need instead of
one’s ability to make a prediction or generate explanations for another’s behavior. The
reliability of tolerance for uncertainty was ranged from .70 to .92.

In order to control for another individual difference, the general communication
Suspicion (Appendix E) (Levine & McComack, 1991a) measure was used. This scale
attempted to measure a predisposition toward viewing the communicative behavior of the
other participant as deceptive. This scale had 14 items, with a seven point Likert type
response format. The reliability is acceptable at .71.

The ethos/credibility scale (Appendix C), created by McCroskey (1999), has

22

presumed three dimensions. Each dimension is measured with six bipolar adjective items
with a seven-point scale (see Appendix C). The six items selected for each dimension
result from the earlier factor-analytic studies on the basis of their consistently high factor
loading and high face validity. The average Alpha reliabilities for these measures are:
Competence .85; Trustworthiness .92 and; Goodwill .92 (McCroskey, 1999). When all
three were scored as a single measure of ethos/credibility, the alpha reliability was .94.

To quantify the PD game, the unit of analysis was the individual. Three games
(see Appendix D) were presented and in each game participants were given a choice
whether or not to cooperate with the other interactant. If a participant circled the
“COOperate” box, that indicated behavioral trust and “1” point was assigned to this
variable. If a participant circled the “Do not Cooperate” box, that indicated a failure to
develop behavioral trust and “0” was assigned. There were three trials in the session
whereby the scores were aggregated with a possible range of 0 to 3 for behavioral trust.

An induction check for the independent variable was conducted to see how the
participants actually perceived the valence of information in terms of its perceived
negativity and deviance. A five-item scale was used to measure the perceived negativity
of the partner’s attributes in terms of its valence, and a four-item scale was used to
measure the perceived deviance of the partner’s attribute. The perceived deviance scale
was created for this research. In a previous study (Yoo, 2003), the reported alpha for the
perceived negativity was .88.

Results

Confirmatory Factor Analysis

Confirrnatory Factor Analyses (Hunter & Gerbing, 1982) were performed to test

23

the validity of the hypothesized factor structure of the items and to assess the reliability
of the measures. A valid set of indicators was attained for the variables of interest. After
deleting one item, the scale measuring the general trust (or the general suspicion scale) in
people had an alpha reliability of .78. All items were retained in the scale that was used
to measure tolerance for uncertainty with an alpha reliability of .85. All items were
retained for perceived competence and perceived trustworthiness from the
ethos/credibility scale by McCroskey, and an alpha reliability of .81 and .88, respectively.
However, one item was deleted for perceived goodwill from the same scale, and it had an
alpha reliability of .80. All items were retained for Clatterbuck’s CLUE7 scale with an
alpha reliability of .93. All items were retained for perceived negativity of the
information given in the session, with an alpha reliability of .87. The perceived deviance
of the information scale reported an alpha reliability of .87 after deleting one item. CFA
showed that errors between predicted correlation and observed correlation were trivial,
and the factor structures passed tests of internal consistency and parallelism.
Induction Check

The induction check for information valence was performed to see whether the
intended effects were produced by the confederates. The induction check took place
immediately after the interaction in order to ensure accurate responses. The participants
were randomly given either “wallet condition” or “restaurant condition,” but analyses
revealed no significant differences between these two conditions at an alpha level of .05.
Thus these two scenarios were combined together for purposes of firrther analysis.

A one-way AN OVA was performed on the ratings of the participants’ perceived

negativity and perceived deviance of the given information about the confederate.

24

 

Participants given the negatively valenced information about the confederate indicated
the highest scores on the perceived negativity scale (M = 2.95, SD = 1.17), followed by
the control information condition (M = 2.69, SD = 1.07). Participants who received a
piece of positive information about the confederate scored the lowest on perceived
negativity scale (M = 1.88, SD = .63). Statistical significance was found in overall mean
differences among these three conditions, F (2, 92) = 10.35, p < .001. Post hoc test
results, using Scheffé’s procedure, indicated that the mean difference of perceived
negativity scale (M0 = -1.07) between the positive information condition and the
negative information condition was statistically significant, p < .001. The mean
difference of the perceived negativity scale (M0 = -.81) between the positive information
condition and the negative information condition was also statistically significant, p < .01.
However, no statistical significance was found in the mean difference between the
negative information condition and the control condition (p = .58).

Statistical significance was found in overall mean differences among these three
conditions in terms of the perceived deviance scale, F (2, 92) = 6.83, p < .001.
Participants who received a piece of negative information about the confederate reported
the highest scores on the perceived deviance scale (M = 2.54, SD = 1.38), followed by
participants in the control information condition (M = 2.24, SD = 1.20). However, the
mean difference between these two conditions (MD = .29) was not statistically significant
(p = .5 8). According to the post hoc comparison using Scheffé’s procedure at p < .05, the
mean difference of the perceive deviance scale (MD = -.99) between the positive
information condition and the negative information condition was Statistically significant,

p < .01. Statistical significance was also found in the mean difference (MD = -.70)

25

between the positive information condition and control condition, p < .05. Therefore,
contrary to the pilot test results, the manipulation for experiment effectively induced the
perceived positivity in participants, while the manipulation for perceived negativity was
less clear.

Hypotheses

The first hypothesis predicted that participants who were exposed to positive
information about their partner (confederate) would reduce the most uncertainty followed
by participants who were exposed to no additional information, and participants exposed
to negative information. The result of a one-way AN OVA indicated that no statistical
significance was found such that different valence of information had no significant
influence on uncertainty reduction, F (2, 92) = 6.83, p = .24, 772 = .03. Mean differences
among these three conditions are trivial (see Table l).

The second hypothesis tested how different valence of information had an
influence on the cognitive state of trust. It was predicted that participants who were
given positive information would score the highest on the ethos/credibility scale
(McCroskey & Teven, 1999). The scale had three dimensions which consisted of
perceived competence, perceived goodwill, and perceived trustworthiness. The result of
a one-way ANOVA indicated that a main effect of information valence on perceived
competence was statistically significant, F (2, 92) = 3.56, p < .05, 772 = .07. Participants
who received a piece of positive information about the partner scored the highest on
perceived competence (M = 5,46, SD = .99), followed by participants who received a
piece of negative information condition (M = 5.18, SD = .72). Participants who received

no additional information about the confederate scored the lowest on perceived

26

competence (M = 4.89, SD = .81). The mean difference between the positive information
condition and the control condition was statistically Significant, p < .05, but the difference
between the positive information condition and the negative information condition (p

= .40), and the difference between the negative information condition and the control
condition (p = .41) were not statistically different. The prediction of direction was
partially supported in that the positive information condition scored highest on perceived
competence, followed by the negative information condition (see Table 1).

Perceived goodwill was analyzed using a one-way AN OVA. The main effect of
information valence on perceived goodwill was statistically significant, F (2, 92) = 3.39,
p < .05, 772 = .07. Similar to the perceived competence dimension, participants given the
positive information about the confederate reported the highest score on perceived
goodwill (M = 4.33, SD = .84), followed by participants who received the negative
information (M = 3.92, SD = 1.15). Again, participants who received no additional
information about the partner expressed the least perceived goodwill toward the partner
(M = 3.73, SD = .75). The mean difference between the positive information condition
and the negative information condition (p = .21), and between the negative condition and
the control condition (p = .71) were not statistically significant. Only the mean difference
between the positive condition and the control condition was statistically significant at p
< .05 (see Table 1).

Perceived trustworthiness was analyzed using a one-way ANOVA. The main
effect of information valence on perceived trustworthiness was statistically significant, F
(2, 92) = 7.12, p < .01, 772 = .13. Participants who received positive information reported

a significantly higher score on perceived trustworthiness toward their partner (M = 5.60,

27

SD = .90) than participants who received no additional information about their partner (M
= 4.83, SD = .79, p < .05) or participants who received negative information about their
partner (M = 4.78, SD = 1.19, p <.001). No statistical significance was found in the mean
difference between the “negative information” condition and the “no information”
condition, (p = .98). The mean pattern was in a predicted direction such that a piece of
positive information most increases the perceived trustworthiness, and a piece of negative
information least increased perceived trustworthiness (see Table 1).

Hypothesis 3 stated that participants who received positive information about
their partner would exhibit more trusting behavior than participants who received no
information, and participants who received negative information. Information valence
had no (or a weak) influence on the behavioral trust, and no statistical significance was
found, F (2, 92) = .02, n.s. (see Table 1).

Path Analysis

To test the proposed model (Figure 2), the least squares criterion was employed
to estimate the parameters. Parameter size as well as the fit of the model was examined.
The correlations employed to estimate the fit of the model parameters are presented in
Table 3, and the path coefficients are presented in Figure 2.

Parameter size was estimated in the path diagram by performing a simple
regression of each endogenous variable onto its causal antecedent. Then model fit was
performed by comparing the estimated parameter size to the reproduced correlations (see
Hunter & Gerbing, 1982 for information on reproducing correlations in path analysis). In
order to say that the model is consistent with the data, two conditions should be met: (1)

the path coefficients should be substantial, and (2) the discrepancies between parameter

28

estimates and reproduced correlations Should be attributable to sampling error. If errors
are larger than what is expected from sampling error, the model is regarded as
inconsistent with the data.

The original model proposed that uncertain level would affect the cognitive state
of trust, which would lead to behavioral trust. Figure 2 illustrates that most of the path
coefficients were substantial, and most of the coefficients were in the direction predicted,
however, not all paths were ample. The coefficient linking the uncertainty reduction
level and perceived competence was within sampling error of zero [.28, P (-.09 S p S .65)
= .95]. Both links between uncertainty reduction level and perceived goodwill [.57, P
(.26 S p S .88) = .95] and uncertainty reduction level and perceived trustworthiness [.55,
P (.26 s p S .84) = .95] were statistically significant. These results indicated that the
level of uncertainty had an influence both on perceived goodwill and perceived
trustworthiness. However, the coefficient linking perceived goodwill and behavioral trust
was within sampling error of zero [.01, P (-.09 S p S .65) = .95], which indicated that
perceived goodwill had no significant effect on behavioral trust. The coefficient linking
perceived competence and behavioral trust was -.13, P (-.68 S p S .42) = .95, which was
not Significantly different from zero as well. Only the coefficient linking perceived
trustworthiness and behavioral trust was statistically significant [.58, P (.07 S p S .99)
= .95]. This indicates that perceived trustworthiness was the only dimension in the
cognitive state of trust which predicted the behavioral trust.

The fit of the overall model was tested for the proposed model. The
discrepancies between obtained and predicted correlations for all bivariate relationships

in the model were examined and many differed significantly from what was expected

29

from sampling error. The largest errors were .46 (between uncertainty reduction level
and behavioral trust), and .27 (between‘perceived competence and behavioral trust).
Overall, the global test for goodness of fit indicated that the data were consistent with the
model, x2(4, N = 32) = 3.22, p = .52, however, a post hoc search for an alternative model
that fit the data was undertaken in order to find a better model.

Evaluation of the Alternative Models

Results indicated the data were consistent with an alternative model (see Figure
3). The coefficient linking perceived trustworthiness and behavioral trust was .35, P (-
.04 S p S .74) = .95 , and the coefficient linking uncertainty reduction level and behavioral
trust was .29, P (-.10 S p S .68) = .95].

The path coefficients in the revised model were ample and were in the direction
predicted. Furthermore, the discrepancies between predicted and obtained correlations
for the unconstrained bivariate correlations demonstrated that the errors were not
significantly different from those predicted from sampling error (the largest error was -
.08). The global test for goodness of fit also indicated that the data were consistent with
the model, {(2, N = 90) = .17, p = .92. Given that the path coefficients were ample, and
that the model and parameter estimates predicted accurately the unconstrained
correlations, the data were consistent with the alternative model.

Discussion

This investigation assessed how different valences of information affect
uncertainty reduction level, cognitive state of trust, and finally, behavioral trust. This
section will discuss results in terms of relationship among constructs and implications of

the study. The discussion of results is followed by (1) extended implications of second

30

order unidimensionality of negativity and deviance, (2) recommendations for scenario
study and its ecological validity, and (3) limitations and directions for future research.
Implications from Proposed Hypotheses

The results of this experiment offer some explanations and implications for
information valence and its effects on trust building in initial interaction. Specifically,
this experiment found that information valence had a significant influence on building the
cognitive state of trust. However, contrary to predictions, information valence had no (or
minimal) influence on building behavioral trust. That is, both correlations between
perceived negativity and behavioral trust (r = -.17), and perceived deviance and
behavioral trust (r = -.05) were in the predicted direction, but not statistically significant.
When looking at only the positively valenced information condition, there was a stronger
correlation between perceived negativity and behavioral trust (r = -.23), and perceived
deviance and behavioral trust (r = -.10). However, it was not statistically significant.

This research challenged Pearce’s claim (1974) that behavioral trust may or may
not be accompanied by building a cognitive state of trust. Further, this research assumed
that a cognitive state of trust should be present prior to behavioral trust. This claim was
partially supported for the positive information condition where the correlation between
perceived trustworthiness and behavioral trust (r = .51) was significant, p <.005. Among
three dimensions of the cognitive state of trust, perceived trustworthiness was the only
dimension that predicted one’s decision in respect to behavioral trust. As predicted, in
the negative information condition, no significant correlation was found between the
cognitive state of trust and behavioral trust. This finding implies that in initial interaction,

it is important to build one’s trustworthiness in order to establish behavioral trust.

31

The results reveal that uncertainty reduction level was closely related with the
cognitive state of trust. The correlation between uncertainty reduction and perceived
competence (r = .25) was significant at p <.05. The correlation between uncertainty
reduction and perceived goodwill (r = .31) was Significant at p <.005, and perceived
trustworthiness (r = .40) was Significant at p <.001. However, no significant relationship
was found between uncertainty reduction level and behavioral trust. The significant
correlation between uncertainty reduction level and behavioral trust (r = .48) was found
only in the positive information condition, p <.01. Therefore, when positive information
was given within the initial interaction with a stranger, there was more chance to reduce
the uncertainty level, and to increase the cognitive state of trust. Under the assumption
that the stranger has positive information about an individual, if the individual
successfully builds perceived trustworthiness, he/she is more likely to build behavioral
trust as an ultimate goal of the interaction. A post hoc contrast analysis was conducted to
see whether having a piece of information, regardless of its valence, had an effect on
cognitive trust, behavioral trust, and uncertainty reduction. The results indicated that
except for perceived competence (t (92) = 2.31, p < .05) of the cognitive state of trust, no
statistical significance was found for information versus no information among perceived
trustworthiness, perceived goodwill, behavioral trust, and uncertainty reduction level.

Tolerance for uncertainty (Appendix B) and general trust were measured in order
to control for individual differences. Tolerance for uncertainty had no significant impact
on any dependent variables, including uncertainty reduction level. The correlation
between general trust and perceived competence was significant, r (95) = .22, p < .05.

General trust was also significantly correlated with perceived goodwill, r (95) = .22, p

32

< .05. These findings indicated that an individual who had higher scores on general trust
scale, tended to perceive their partners to be more competent and to have more goodwill

toward them. General trust was positively correlated with perceived trustworthiness, but
no statistical Significance was found, r (95) = .16, n.s.

However, situational variables, such as perceived negativity and deviance, tended
to have stronger correlations with the cognitive state of trust than personality variables,
such as tolerance for uncertainty and general trust in people. For instance, perceived
negativity was negatively correlated with perceived competence, (r = -.43), perceived
goodwill (r = -.35), and perceived trustworthiness (r = -.58). These correlations were all
significant at a p-level of .001. Perceived deviance also indicated the similar results that
perceived deviance was negatively correlated with perceived competence, (r = -.29), with
perceived goodwill (r = -.32), and with perceived trustworthiness (r = -.39). All of these
correlations were also significant at p-level of .005. However, neither personality
variables nor Situational variables had significant correlations with behavioral trust,
except for in the positive information condition.

As the path model suggested both perceived trustworthiness and uncertainty
reduction level were significant predictors for behavioral trust only when an individual
was exposed to a piece of positive information about the other partner. When people
heard a piece of negative information about their partner, no constructs predicted the
behavioral trust. Perceived negativity and deviance were negatively correlated with the
cognitive state of trust, but no significant relationship was found between the cognitive
state of trust and behavioral trust, or uncertainty reduction level and behavioral trust.

A few issues still remain unsolved and should serve as a future mission. The first

33

issue is that when controlling for perceived trustworthiness, the direct path between
uncertainty reduction and behavioral trust disappears, which suggests that there might be
the third variable that affects both uncertainty reduction and behavioral trust. The second
issue concerns the time order criteria of the causal relationship. This study was an
experimental study, and made an assumption that the cognitive state of trust came prior to
behavioral trust. Also according to the study design, participants had to fill out the scales
that were asking for ratings of partner (confederate)’s perceived competence, perceived
goodwill, and perceived trustworthiness before they played the Prisoner’s dilemma game
which was the indicator of behavioral trust. The question should be asked in future
research whether this assumption of time ordering can be sufficiently justified.
Implications for Dimensionality of Negativity and Deviance

This research originally attempted to measure two different constructs of
information valence: perceived negativity and perceived deviance. The items in the
perceived negativity scale focused on how the given information was good/bad or
positive/negative, while the perceived deviance scale focused on how the given
information was normal/abnormal or ordinary/deviant. The alpha reliability for the
perceived negativity scale and the perceived deviance scale were both .87 when treating
these two as separate constructs. The two largest errors between the predicted
correlations and the observed correlations were .30 and .15.

However, when these two constructs were merged and considered to be a
unidimensional construct, the alpha reliability increased to .92. The errors between the
predicted correlation and the observed correlation appeared to be trivial, that the largest

errors were .24 and .20. The factor loadings of the items in the perceived negativity scale

were also highly loaded on the perceived deviance scale, and the factor loadings of items
on the perceived deviance scale were also highly loaded on the perceived negativity scale.
These two constructs of the perceived negativity and the perceived deviance had similar
relationships with other constructs. For instance, both constructs were negatively
correlated with general trust (or general suspicion) in people (r = -.26, p <.05; r = -.31, p
<.01); with perceived competence (r = -.43, p <.001; r = -.29, p <.001); with perceived
goodwill (r = -.35, p <.001; r = -.32, p <.001); and perceived trustworthiness (r = -.58,
p <.001; r = -.39, p <.001). Other constructs, including tolerance for the uncertainty,
uncertainty reduction level, and behavioral trust, were not statistically significantly
correlated with either of these two constructs. Therefore, the CF A suggests that these
two constructs, perceived negativity and perceived deviance, appeared to be under a
higher order factor, such as valence of information or evaluation of information.
Recommendations for Scenario Study and its Ecological Validity

A common criticism of hypothetical scenarios may also apply to the results of
this study, as the use of hypothetical scenarios may cause problems in this line of
research. The results of this study questioned the validity of the hypothetical scenario
study. This study used various hypothetical scenarios as a pretest in order to come up
with a valid induction for information valence for the actual experiment. The same
negative information was measured in both the hypothetical scenario study and the actual
face-to-face experiment. In the pretest using the hypothetical scenario, inductions for
information valence (positively valenced and negatively valenced information conditions)
were more successful in terms of magnitude of mean differences for perceived negativity

and deviance than the actual experimental inductions. For instance, when respondents

35

read the hypothetical scenario (restaurant scenario) of the negatively valenced condition,
they reported that the information about their blind partner (or group member) was
perceived to be Significantly more negative (M = 5.17 , SD = 1.35) than when they
received no information about their partner (M = 2.81, SD = .99, t(42) = 11.34, p < .001).
Similar results were found in terms of perceived deviance in that respondents in
negatively valenced information condition reported significantly higher perceived
deviance (M = 4.64, SD = 1.12) than respondents who were in the control condition (M =
2.33, SD = 1.13, t(4l) = 13.35,p < .001). The mean patterns specified that both
perceived negativity and deviance exceeded significantly the midpoint of the scale, which
soundly proved that the induction for negatively valenced information was successful.

However, the actual experiment data consistently showed that the participants in
the negatively valenced information condition were not significantly different from the
participants in the control condition, in terms of either perceived negativity (MD = .26) or
perceived deviance (MD = .29). In fact, means for perceived negativity (M = 2.94, SD =
1.17) and perceived deviance (M = 2.53, SD = 1.38) in a negatively valenced condition,
although the highest scores among three conditions, did not exceed the midpoint of the
scale. Therefore, participants who read negative information about their partner
perceived more negativity and deviance in respect to the given information to participants
who interacted face-to-face and received information directly from their partner.

Effects sizes are noticeably different between the hypothetical scenario study and
the actual experiment. For perceived negativity, the effect size for the scenario study was

772 = .67, while the effect size for the actual experiment was 172 = .18. Similarly, for

perceived deviance, the effect size for the scenario was 172 = .49, while the effect size for

36

the actual experiment was 772 = .13. These results suggest that when individuals directly
interact with a stranger, they tend to evaluate their partner to be more positive/ordinary
(or much less negative/deviant) than when they indirectly interact with the stranger via a
print medium. These results indicated that continuing reliance on hypothetical scenarios
and responses in this line of research may result in exaggerated findings that are
discrepant from reality. These findings from the hypothetical scenario study may not
accurately reflect on how participants really respond to various actual situational
influences.
Limitations and Future Research

Three limitations of the results of this investigation must be addressed. The first
caveat involves the Operationalization of behavioral trust. Deutsch (1958) was the first
scholar who employed the Prisoner’s Dilemma game in order to operationalize
behavioral trust. He found that communicating with a partner increased the individual’s
confidence whereby his/her trust would be fulfilled and also used to elicit trustworthy or
responsible behavior. In his study, he induced the experimental situation using the
Prisoner’s Dilemma game in a way that participants should bet their own money. Thus,
if they lost to their partner, it was an actual loss of the money. This strong induction
allowed for researchers to measure a highly valid set of behavioral trust. However, this
investigation used a weak induction in that if participants won the highest score on these
three trial-games, they would receive a chance at the $50.00 gift certificate from the
experimenter. Therefore, their loss was not directly related to their own monetary value,
which might be problematic in terms of inducing valid behavioral trust. In other words,

perceived risk was not necessarily (or sufficiently) induced in this investigation.

37

 

Perceived risk is an essential construct in terms of exploring behavioral trust. Behavioral
trust should involve the voluntary action of making oneself vulnerable to the decisions of
others, while cognitive state of trust only concerns one’s perception about the other
(Pearce, 1974). In cognitive state of trust, unlike behavioral trust, one is not likely to risk
anything, thus, one’s vulnerability is minimal and limited. Under the assumption that the
induction for information valence was successful, a weak induction for behavioral trust
using the Prisoner’s Dilemma might have undermined the effect on information valence
on behavioral trust.

The ecological validity issue should be noted as well. The laboratory study
appears to control for any external factors that might interfere with the internal validity,
while the ecological validity concerns the extent to which findings from the laboratory
can be generalized to reality. It is inevitable that studies in a lab always involve a trade-
off between control issues and ecological validity. As previously mentioned, the gap
between the results in hypothetical scenario study and experimental study showed that
people respond differently in each. A further research question should consider how
people would react to situations in the “real world.” AS the results of this investigation
already suggested, the negative effect might be mitigated when there is actual face-to-
face interaction.

Also, it should be noted that the independent variable might have been
confounded within the experimental design. Especially, when the confederate gave the
participant the negative information about himself/herself, participants might have
perceived “negative information giving” as “willingness to admit one’s fault” rather than

the originally intended effect of “perceived negativity.” Thus, participants might have

38

not perceived negativity in the information given by the confederate. In order to measure
a “true” assessment of the stimuli, “willingness to admit one’s fault” should be controlled
statistically, or the researcher Should use an experimental design in which a third person
presents different valenced information about the confederate.

The current study allows for future research to determine the causes of
discrepancies between the hypothetical scenario study and the actual experiment. More
specifically, future research should attempt to explain why the negativity effect plays a
more significant role in the hypothetical situation than in the actual experiment. Future
research should also investigate how to restore damaged images when a partner is already
aware of negative information about them. This study showed that in the actual
experimental situation, the same piece of negative information was perceived to be
significantly less negative and deviant than in the hypothetical scenario situation. This
suggests that when having face-to-face interaction with a stranger for the first time,
people in general tend to have a positive disposition toward others. Therefore, future
research should incorporate a positivity bias and a negativity effect in order to determine
how information valence has a significant impact on impression formation in the initial

interaction with a stranger.

39

 

Table 1. Means and (Standard Deviations) for Each Dependent Variables Broken Down

By Information Valence Type

 

 

 

 

 

 

 

 

 

 

 

 

 

Information Valence Range Negative No additional Positive
information information information

Perceived negativity 1-7 2.95 (1.17) 2.69 (1.07) 1.88 (.63)
Perceived deviance 1-7 2.54 (1.38) 2.24 (1.20) 1.54 (.60)
Uncertainty Reduction 0-100 40.98 (19.85) 32.52 (21.42) 40.60 (24.37)
Perceived competence 1-7 5.18 (.72) 4.89 (.81) 5.46 (.99)
Perceived goodwill 1-7 3.92 (1.15) 3.73 (.75) 4.33 (.84)
Perceived trustworthiness 1-7 4.78 (1.19) 4.83 (.79) 5.60 (.90)
Behavioral trust (3 trials) 0-3 .73 (.33) .73 (.28) .72 (.25)
Behavioral trust (first trial) 0-1 .85 (.36) .90 (.31) .81 (.40)

 

 

40

 

Table 2. Zero Order Correlations among Variables

(1) Zero order correlations among the variables in positively valenced information

condition (N = 32)

1 2 3 4 5
Respondents (n = 32)
1. Perceived competence 1.00
2. Perceived goodwill .24 1.00
3. Perceived trustworthiness .61** .43* 1.00
4. Uncertainty reduction .28 .57 .55 1.00
5. Behavioral trust .23 .23 .51** .48** 1.00

Note. * p < .05, ** p < .01, N=32.

(2) Zero order correlations among the variables in negatively valenced information

condition (N = 33)

1 2 3 4 5
Respondents (n = 33)
1. Perceived competence 1.00
2. Perceived goodwill .30 1.00
3. Perceived trustworthiness .40* .41 * 1.00
4. Uncertainty reduction .31 .19 .23 1.00
5. Behavioral trust .05 -.30 .16 -.02 1.00

Note. * p < .05, ** p < .01, N=33.

41

Figure 1. PD Game

 

 

A B
Person H does not Person 11 does
confess confess

 

X Person I does not
confess.

Person I: 1 year
Person II: 1 year

Person 1:10 year
Person 11: go free

 

Y Person I does
confesses.

 

 

 

Person I: go free
Person II: 10 year

 

Person I: 9 year
Person II: 9JIear

 

42

 

Figure 2. Initial Path-Analytic Model

Positively valenced information condition

.28

 

Uncertainty , 7
Level

 

 

.55

N = 32, x2 (4) = 3.22, p = .51

General Trust

 

 

Perceived Competence

 

 

 

 

Perceived Goodwill

-.l3

 

.01 Behavioral

 

 

 

 

 

Perceived Trustworthiness

 

 

43

 

 

 

Trust
KL

Figure 3. Revised Path-Analytic Model

Positively valenced information condition

 

Perceived
Trustworthiness

 

.35

 

 

\

 

 

Uncertainty
Reduction

 

 

 

 

Behavioral
Trust

 

N= 32, x2 (2) = .17, p= .92

44

 

Appendix A. CL7 Attributional Confidence Scale (Clatterbuck, 1979)

revised version (Gudykunst & Nishida, 1986)

Instructions: People vary in the degree to which they can predict how other people
behave and think. Please answer each of the following questions with respect to your
ability to predict selected aspects of the behavior of the person you answered the previous
question about. Answer each question using a scale from zero (0) to one hundred (100).
If you would have to make a total guess about the person’s behavior of feelings you

should answer “0”; if you have total certainty about the other person’s behavior you
should answer “100.” Feel free to use any number between 0 and 100.

_ 1. How confident are you of your general ability to predict how he/she will behave?
_ 2. How certain are you that he/She likes you?

__ 3. How accurate are you at predicting the values he/she holds?

_ 4. How accurate are you at predicting his/her attitudes?

_ 5. How well can you predict his/her feelings and emotions?

_ 6. How much can you empathize with (share) the way he/she feels about
himself/herself?

_ 7. How well do you know him/her?
_ 8. How certain are you of his/her background?

_ 9. How certain are you that he/she will behave in a socially appropriate way when
this is important?

__ 10. How certain are you that he/she can understand your feelings when you do not
verbally express them?

_ 11. How certain are you that you understand what this person means when you
communicate?

_ 12. How confident are you that this person will make allowances for you when you
communicate?

45

Appendix B. Tolerance for Uncertainty scale

Instructions: Please indicate your attitude about the world noted below by indicating the
appropriate number.

Notatall 1 2 3 4 5 6 7 8 9 Verymuch

_ 1. How confident do you feel you need to be in your general ability to predict how
the person will behave in general?

_ 2. How certain do you need to be about how well the person likes you?

_ 3. How accurate do you think you need to be at predicting the values the person
holds?

_ 4. How accurate do you think you need to be at predicting the person’s attitudes?

__ 5. How well do you think you need to be able to predict the person’s feelings and
emotions?

_ 6. How much do you think you need to be able to emphasize (share) the way the
person feels about him/her as a person?

_ 7. How well do you think you need to know the person?

_ 8. How well do you think you need to understand the person?

46

Appendix C. Measure of Ethos/Credibility

Instructions: Please indicate your impression of your partner as noted below by circling
the appropriate number between the pairs of adjectives below. The closer the number is
to an adjective, the more certain you are of your evaluation.

Competence
Intelligent
Untrained
Inexpert
Informed
Incompetent

Bright

Mil

Cares about me

Has my interests at heart
Self-centered
Concerned with me
Insensitive

Not understanding

Trustworthiness
Honest
Untrustworthy
Honorable
Moral
Unethical
Phoney

Hp—ap—Ay—dp—ap—o p—Ap—aHHp—Ay—a

Hp—ip—fip—np—dp—A

NNNNNN

NNNNNN

NNNNNN

wwwwww

wwwwww
h-h-h-b-bh
MMMMMM

wwwuww
A-b-b-b-h-D-
MMMMMM

Ji-b-«b-h-b-h

MLI'ILIILI’IMLII

O\O\O\O\O\O\

O\O\O\O\O\O\

O\O\O\O\O\O\

\l\l\l\l\l\l \l\l\)\l\l\)

\l\l\l\l\l\l

Unintelligent*
Trained
Expert
Uninformed’“
Competent
Stupid*

Doesn’t care about me
Doesn’t have my interest at heart
Not self-centered*
Unconcemed with me
Sensitive*

Understanding*

Dishonest*
Trustworthy
Dishonorable*
Immoral*
Ethical
Genuine

 

* reversed items

47

Appendix D. Prisoner’s Dilemma Game

Game 1:

 

Your partner cooperates

Your partner does not
cooperate

 

You cooperate

Both receive 3 pts.

Your partner receives 5 pts.
You receive 2 pt.

 

You do not
cooperate

 

 

You receive 5 pts.
Your partner receives 2pt.

 

Both receive no pts.

 

What is your final choice?

Cooperate

Do not Cooperate

Please wait until the experimenter guides you to the next step of the experiment.

 

Game 2:

 

Your partner cooperates

Your partner does not
cooperate

 

You cooperate

Both receive 3 pts.

Your partner receives 7 pts.
You receive 1 pt.

 

You do not
comrate

 

 

You receive 7 pts.
Your partner receives lpt.

 

Both receive no pts.

 

What is your final choice? Cooperate

Do not Cooperate

Please wait until the experimenter guides you to the next step of the experiment.

 

Game 3:

 

Your partner cooperates

Your partner does not
cooperate

 

You cooperate

Both receive 3 pts.

Your partner receives 10 pts.
You receive -5 pt.

 

You do not
cooperate

 

 

You receive 10 pts.
Your partner receives -5pt.

 

Both receive -7 pts.

 

What is your final choice? Cooperate

Do not Cooperate

Please wait until the experimenter guides you to the next step of the experiment.

48

 

 

 

Appendix B. General Suspicion Scale

Instructions: Please indicate your attitude about the world noted below by indicating the
appropriate number.

Strongly Disagree 1 2 3 4 5 6 7 Strongly Agree

_ 1. I often feel as if people aren’t being completely trustful with me.
_ 2. Most people only tell you what they think you want to hear.

_ 3. When I am in a conversation with someone, I frequently wonder whether they are
really telling me the truth.

_ 4. People rarely tell you what they’re really thinking.

__ 5. When I first meet someone, I assume that they are probably lying to me about
some things. '

_ 6. Most peOple are basically honest.
_ 7. Anyone who completely trusts someone else is asking for trouble.

_ 8. When I ask a stranger for directions, I frequently wonder whether they are being
truthful.

_ 9. When I am talking to others, I tend to believe what they say.
_ 10. People seldom lie to me.

__ 11. Most people follow the saying “honesty is the best policy.”

49

Appendix E: Manipulation Check

Instructions: Please indicate your attitude about your partner below by indicating the

appropriate number.

I think the information I received about my partner during the interaction was

- Perceived Negativity

Negative

Something I would rate
unfavorably
Something I like
Something that I would
respect

Bad

- Perceived Deviance

Deviant
Inappropriate
Normal

Desirable

N

NNNN

b)

DJLAJUJ

fish-5%
UIUIKIIQII

50

O’\O\O\O\

\)\I\I\l

Positive

Something I would rate
favorably

Something I don’t like

Something that I would not
respect
Good

Ordinary
Appropriate
Abnormal
Undesirable

References

Altman, I., & Taylor, D. A. (1973). Social penetration: The development of interpersonal
relationships. New York: Holt: Rinehart and Winston.

Anderson, N. H. (1965). Averaging versus adding as a stimulus - combination rule in
impression formation. Journal of Experimental Psychology, 70, 394-400.

Barber, B. (1983). The logic and limits of trust. New Brunswick, NJ: Rutgers University
Press.

Bem, D. J. (1970). Beliefls, Attitudes, and Human Affairs. Pacific Grove, CA:
Brooks/Cole.

Berger, C. R. (1986). Uncertain outcome values in predicted relationships: Uncertainty
reduction theory then and now. Human Communication Research, 13, 34-38.

Berger, C. R., & Calabrese, R. J. (1975). Some explorations in initial interaction and
beyond: Toward a development theory of interpersonal communication. Human
Communication Research, 1, 99-112.

Berger, C. R., & Gudykunst, W. B. (1991). Uncertainty and Communication. In B.
Dervin & M. J. Voight (Eds), Progress in Communication Science (pp. 21-66).
Norwood: NJ: ABELEX Publishing Corporation.

Berlo, D. H., Lemert, J. B., & Mertz, R. J. (1969). Dimensions for evaluating the
acceptability of message sources. Public Opinion Quarterly, 33, 563-576.

Brinbaum, M. H. (1973). Morality Judgment: Test of an averaging model with
differential weights. Journal of Experimental Psychology, 99, 395-399.

Canary, D. J ., & Cupach, W. R. (1988). Relational and episodic characteristics associated
with conflict tactics. Journal of Social and Personal Relationships, 5, 305-325.

Clatterbuck, G. W. (1979). Attributional confidence and uncertainty in initial interaction.
Human Communication Research, 4, 147-157.

Deutsch, M. (1958). Trust and suspicion. Journal of Conflict Resolution, 2, 265-279.

Douglas, W. (1990). Uncertainty, Information-seeking, and liking during initial
interaction. Western Journal of Speech Communication, 54, 66-81.

Fiske, S. T. (1980). Attention and weight in person perception: The impact of negative
and extreme behavior. Journal of Personality and Social Psychology, 38, 889-906.

Giffin, K. (1967). The contribution of studies of source credibility to a theory of

51

interpersonal trust in the communication process. Psychological Bulletin, 68, 104-
120.

Gudykunst, W. B. (1985). The influence of cultural similarity, type of relationship, and
self-monitoring on uncertainty reduction processes. Communication Monographs,
52, 203-217.

Gudykunst, W. B., & Kim, Y. Y. (1997). Communicating with strangers: An approach to
intercultural communication. New York: McGraw-Hill.

Gudykunst, W. B., & Nishida, T. (1986). Attributional confidence in low- and high-
context cultures. Human Communication Research, 12, 525-549.

Hamilton, D. L., & Zanna, M. P. (1972). Differential weighting of favorable and
unfavorable attributes in impression formation. Journal of Experimental Research
in Personality, 6, 204-212.

Harvey, J. H., Wells, G. L., & Alvarez, M. D. (1978). Attribution in the context of
conflict and separation in close relationships. In J. H. Harvey, W. Ickes & R. F.
Kidd (Eds.), New Directions in Attribution Research (Vol. 2, pp. 235-260).
Hillsdale, NJ: Lawrence Erlbaum.

Holmes, J. G., & Rempel, J. K. (1989). Trust in close relationships. In C. Hendrick (Ed.),
Review of Personality and Social Psychology: Close Relationships (Vol. 10, pp.
187-219). Newbury Park: Sage.

Hosmer, L. (1995). Trust: The connecting link between organizational theory and
philosophical ethics. Academy of Management Review, 20, 379-403.

Hunter, J ., & Gerbing, D. (1982). Unidimensional measurement, second order factor
analysis, and causal models. Research in Organizational Behavior, 4, 167-320.

J ohnson-George, C., & Swap, W. (1983). Measurement of specific interpersonal trust:
Construction and validation of a scale to assess trust in a specific other. Journal of
Personality and Social Psychology, 43, 1306-1317.

Kellermann, K. (1989). The negativity effects in interaction: It's all in your point of View.
Human Communication Research, 16, 147-183.

Kellermann, K., & Reynolds, R (1990). When ignorance is bliss: The role of motivation
to reduce uncertainty in Uncertainty Reduction Theory. Human Communication
Research, 1 7(1), 5-75.

Larzelere, R. E., & Huston, T. L. (1980). The dyadic trust scale: Toward understanding

interpersonal trust in close relationships. Journal of Marriage and the Family, 42,
595-604.

52

Levine, T. R., & McComack, S. A. (1991a). The dark Side of trust: Conceptualizing and
measuring types of communicative suspicion. Communication Quarterly, 49(4),
325-340. ‘

Lewicki, R. J ., & Bunker, B. B. (1995). Trust in relationships: A model of trust
development and decline. In B. B. Bunker & I. Z. Rubin (Eds.), Conflict,
Cooperation and Justice. San Francisco: Jossey-Bass.

Mayer, R., Davis, J ., & Schoorman, F. (1995). An integrative model of organizational
trust. Academy of Management Review, 20, 709-734.

McCroskey, J. C., & Teven, J. J. (1999). Goodwill: A reexamination of the construct and
its measurement. Communication Monographs, 66, 90-103.

McCrosky, J. C. (1966). Scales for the measurement of ethos. Speech Monographs, 33,
65-72.

Miller, F. E., & Rogers, E. (1987). Relational dimensions of interpersonal dynamics. In
M. E. Roloff & G. R. Miller (Eds.), Interpersonal Processes: New Dimensions in
Communication Research. Newbury Park: Sage.

Oden, G. C., & Anderson, N. H. (1971). Differential weighting in integration theory.
Journal of Experimental Psychology, 89, 152-161.

O'Keefe, D. J. (1990). Persuasion: Theory and Research. Thousand Oaks: Sage.

Pearce, B. (1974). Trust in interpersonal communication. Speech Monographs, 41, 236-
244.

Planalp, S., & Honeycutt, J. M. (1985). Events that increase uncertainty in personal
relationships. Human Communication Research, 11, 593-604.

Reeder, G. D., & Coovert, M. D. (1986). Revising an impression of morality. Social
Cognition, 4, 1-17.

Rempel, J. K., Ross, M., & Holmes, J. G. (2001). Trust and communicated attributions in
close relationships. Journal of Personality and Social Psychology, 81, 57-64.

Richey, H. W., McClelland, L., & Shimkunas, A. M. (1967). Relative influence of
positive and negative information in impression formation and persistence.
Journal of Personality and Social Psycholog, 6, 322-327.

Richey, M. H., Koenigs, R. J ., Richey, H. W., & Fortin, R. (1975). Negative salience in

impressions of character: Effects of unequal proportions of positive and negative
information. The Journal of Social Psychology, 97, 233-241.

53

Richmond, C. (1985). Background for the uninitiated: Paradoxes of Rationality and
Cooperation Prisoner's Dilemma and Newcomb's Problem. Vancouver: The
University of British Columbia Press. ‘

Rousseau, D. M., Sitkin, S. B., Burt, R. S., & Camerer, C. (1998). Not so different after
all: A cross-discipline view of trust. Academy of Management Review, 23(3), 393-
404.

Scheidel, T. M. (1977). Evidence varies with phases of inquiry. Western Journal of
Speech Communication, 41, 20-31.

Sheppard, B. H., & Sherman, D. M. (1998). The grammars of trust: A model and general
implications. Academy of Management Review, 23, 422-437.

Sunnafrank, M. (1986). Predicted outcome value during initial interactions: A

reformulation of uncertainty reduction theory. Human Communication Research,
13, 3-33.

Sunnafrank, M. (1990). Predicted Outcome Value and Uncertainty Reduction Theories:
A test of competing perspectives. Human Communication Research, 1 7, 76-103.

Van der Pli gt, J ., & Eiser, J. R. (1980). Negativity and descriptive extremity in
impression formation. European Journal of Psychology, 10, 415-419.

VanLear, C. A., & Trujillo, N. (1986). On becoming acquainted: A longitudinal study of
social judgment processes. Journal of Social and Personal Relationships, 3, 375-
392.

Wheeless, L. R., & Gtotz, J. (1977). The measurement of trust and its relationship to self-
disclosure. Human Communication Research, 3(3), 250-257.

Yoo, J. (2003). Uncertainty reduction and information valence: Tests of uncertainty

reduction theory, predicted outcome value, And an alternative explanation?
Unpublished manuscript, Michigan State University - East Lansing.

54

   

lllglllngtjlijj111311