THEM
l

N 0'
f. ,"

a; ,

6:341 6%]

This is to certify that the
thesis entitled

Persuasion through the Revelation of Self-lncriminating
Information: An Examination of Inconsistencies
in the Stealing Thunder Literature

presented by

Michael Ryan Kotowski

has been accepted towards fulfillment
of the requirements for the

Master of Arts degree in Communication

 

 

Q [I] " a
.» {/0 . \J «a,
i/

Major Professor’s Signature
3:1:1/L' H

Date

MSU is an Affirmative Action/Equal Opportunity Institution

 

 

 

 

UBRARY '
Michigan State
University

 

 

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

 

HQ? 2 ‘9 “I?“

 

 

 

 

 

 

 

 

 

 

 

 

 

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

 

 

PERSUASION THROUGH THE REVELATION OF SELF-INCRIMINATING
INFORMATION: AN EXAMINATION OF INCONSISTENCIES
IN THE STEALING THUNDER LITERATURE
By

Michael Ryan Kotowski

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF ARTS
Department of Communication

2004

ABSTRACT
PERSUASION THROUGH THE REVELATION OF SELF-INCRIMINATING
INFORMATION: AN EXAMINATION OF INCONSISTENCIES
IN THE STEALING THUNDER LITERATURE
By

Michael Ryan Kotowski

Past research on the outcomes of stealing thunder, or revealing incriminating
information about oneself as part of a persuasive message, commonly resulted in
inconsistent findings. This study attempts to reconcile some of the inconsistencies by
varying stealing thunder messages in a crossed refutation, awareness, and communication
context 2 (refutation) X 2 (aware) X 3 (context) mixed design with two offset controls
(no information control and thunder control), with context being the repeated measure.
Participants were randomly assigned to one of the six experimental conditions. They
read three vignettes describing a persuasive situation where the agent employs stealing
thunder, after which they responded to measures of perceived persuasiveness,
information valence, and perceived source credibility. Findings were not consistent with
the proposed hypotheses. The findings do, however, suggest some insight into the

cognitive processes that may be taking place in the target of a stealing thunder message.

ACKNOWLEDGEMENTS

First and foremost, I would like to acknowledge the guidance of my major
professor, Dr. Franklin J. Boster. Over the past two and one half years since I entered the
graduate program his standards and values have made an enormous positive impact on
both my intellectual and personal growth. Although his comments and suggestions from
time to time have caused me to doubt my ability, every single one has nonetheless been
invaluable. Dr. Boster has been a model of academic scholarship and without his untiring
assistance I would not be where I am today.

Committee member, Dr. Timothy R. Levine, was the source for many valuable
discussions about and critiques of my conceptual groundwork for this thesis when I
brought the idea to him after UCRIHS barred work from continuing on my original thesis
(which one day will continue when a just review board is found). Committee member,
Dr. Toby TenEyck, also deserves acknowledgement for always providing me material
and experience that have caused me to think about social phenomena from perspectives
that I might not have considered.

A thank you is in order for Dr. Kip Williams who was gracious enough to provide
me with several of the unpublished manuscripts that were reviewed for this thesis. A
special thanks must also go out to UCRIHS. I will forever feel honored by the committee
that served during the time period of November 8"‘, 2002 through February 3rd, 2003 who
felt my original thesis was worthy of being only the second study in 10 years that had a
design so unethical that work on it was barred from continuing.

Finally, I would like to thank my entire family and especially Erin; I am grateful

for the constant support and encouragement.

iii

TABLE OF CONTENTS

LIST OF TABLES .................................................................................... v
LIST OF FIGURES ................................................................................. vi
INTRODUCTION .................................................................................. 1
METHOD .......................................................................................... 13
Design ...................................................................................... 13
Participants ................................................................................. 13
Procedure .................................................................................. 13
Interaction Scenarios ...................................................................... l4
Perceived Persuasiveness Instrument .................................................. 15
Source Credibility Instrument ............................................................ 15
Information Evaluation Instrument ....................................................... 16
Induction Check Instrument ............................................................. 16
Demographics Instrument ............................................................... 16
RESULTS ............................................................................................ 17
Source Credibility Measurement Model ............................................... 17
Information Evaluation Measurement Model ......................................... 17
Induction Checks ......................................................................... 19
Hypothesis Tests .......................................................................... 23
DISCUSSION ...................................................................................... 33
APPENDICIES
Appendix A: Interaction Scenarios — Legal Context. . . .. .................................. 44
Appendix B: Interaction Scenarios — Interpersonal Context. . . .. ...................... 48
Appendix C: Interaction Scenarios — Relational Context. . . . . . . . . . . . . . . . . ...........52
Appendix D: Persuasiveness Instrument ............................................... 55
Appendix E: Source Credibility Instrument. 56
Appendix F: Information Evaluation Instrument.................................. ....59
Appendix G: Induction Check Instrument60
Appendix H: Demographics Instrument61
REFERENCES ....................................................................................... 63

iv

Table 1

Table 2

Table 3

Table 4

Table 5

Table 6

Table 7

Table 8

Table 9

Table 10

LIST OF TABLES

Hypothesized Additive Relationship between Trustworthiness

and Expertise Inductions and Perceived Source Credibility. . . . . . . . . ..

Hypothesized Additive Relationship between Trustworthiness

and Expertise Inductions and Information Valence Change. . . . . . . . . . .

Hypothesized Additive Relationship between Trustworthiness

and Expertise Inductions Resulting Perceived Persuasiveness. . . . .

Factor Loadings for the Unidimensional Source Credibility

Solution by Context .........................................................

Factor Loadings for the Unidimensional Information Evaluation

Solution by Context .........................................................

Observed Relationship between Experimental Inductions and

Perceived Source Credibility ...............................................

Observed Relationship between Experimental Inductions and

Information Valence Judgment ............................................

Observed Relationship between Experimental Inductions and

Perceived Persuasiveness ...................................................

Dunnett’s T Results Comparing Source Credibility
Mean Differences between the No Information Control and

Other Groups in the Interpersonal Context

Dunnett’s T Results Comparing Perceived Persuasiveness
Mean Differences between the No Information Control and

Other Groups in the Interpersonal Context.

11

12

12

18

19

25

27

29

31

31

LIST OF FIGURES

Figure 1 Proposed model: Influence of stealing thunder, source credibility,
and information valence on persuasion .......................................... 9

vi

INTRODUCTION

A common theme in much of the literature on persuasion strategies is the apparent
importance of portraying the influence agent in a positive light. For example, the
persuasive advantage of creating an early positive impression on the target has been
repeatedly documented in the literature (Kelley, 1950; O’Keefe, 2002). There are,
however, circumstances in which it may be impossible for the agent to be portrayed in a
positive light. Picture a defendant on trial for manslaughter, or a major oil company that
just received word an oil tanker of theirs spilled its contents off the coast of Alaska, or the
ex-convict trying to make an entrance back into society. A common theme among these
scenarios is that the sources want to convince someone, but all run the risk of being
portrayed in a negative light before they attempt to persuade the prosecutor, the media, or
the parole officer. Situations like these induced Williams, Bourgeois, and Croyle (1993)
to ask the question of whether it could ever be adaptive to reveal negative or damaging
information about oneself as part of a persuasive message before an outside source has
the chance to reveal the negative information.

Williams et al.’s (1993) question subsequently gave rise to the explication of
stealing thunder as a persuasion strategy. Specifically, whereas thunder involves the
revelation of negative information about the agent by an outside source, stealing thunder
involves the disclosure of negative information about oneself before it can be revealed
through some other means (Williams et al., 1993), such as an outside agent. For
example, juries have been found to give more favorable sentences to defendants whose
lawyers stole thunder and revealed negative information about their client before the

prosecution had the opportunity to do so than lawyers who did not steal thunder (Dolnik,

Case, & Williams, 2003; McElhaney, 1987; Williams et al., 1993). Furthermore,
Williams & Dolnik (2001) present unpublished findings by Ondrus & Williams (1995)
demonstrating when a political candidate stole thunder by admitting participation in a
scandal, voters’ willingness to vote for the candidate rose to a level almost as if there had
been no scandal at all. The literature suggests the persuasive effects of stealing thunder
are particularly robust, the effect being seem across context (Dolnik et al., 2003; Ondrus
et al., 1995; Zablocki, 1996) and variation in message features (Dolnik et al., 2003;
Williams et al., 1993). Nevertheless, the literature demonstrates some patterned variation
in stealing thunder’s effectiveness. These patterns suggest factors which mediate the
effectiveness of the strategy.

Williams has produced a program of research attempting to address the causes of
the variation in stealing thunder’s effectiveness (Williams et al., 2001), however, upon
reviewing the literature there appear to be several uncontrolled factors. For example, the
effectiveness of stealing thunder exhibits considerable variation depending on the context
in which it is used. Specifically, in an unpublished paper reported by Williams et a1.
(2001) the stealing thunder effect has been shown to be largest in relational
communication settings (Zablocki, 1996) and smallest in legal settings (Williams et al.,
1993), while interpersonal settings produce intermediate effects (Ondrus et al., 1995).
When the methods of measuring stealing thunder’s effectiveness are considered,
however, the simple claim there are context effects becomes tenuous. The Zablocki
(1996) study used a compliance measure as the dependent variable, Ondrus et a1. (1995)

employed a measure of perceived persuasiveness, and participants in Williams et a1.

(1993) reported dichotomous guilt judgments. As a result, the effects of context are
confounded with the method of measurement.

A closer examination of the literature reveals the study of stealing thunder has
been restricted, in the main, to communication in formal settings such as the courtroom
(Dolnik etal., 2003; Williams et al., 1993) and the political arena (Ondrus et al., 1995;
White & Williams, 1998). These studies produce a common pattern, namely, the agent is
perceived as being most persuasive when there is no revelation of negative information
(control). The agent’s persuasiveness in the control condition is followed by
persuasiveness in a condition in which they reveal the information before an outside
source has the chance (stealing thunder). Finally, when an outside source reveals the
negative information before the agent has the chance (thunder) the agent is reported as
being the least persuasive. Williams et al. (2001) mention one study (Zablocki, 1996)
testing stealing thunder in a relational context in which the agent attempted to secure a
date, the negative information being that the agent was recently diagnosed with a curable
STD. Within the Zablocki (1996) study the agent in the stealing thunder condition was
more persuasive (70% agreed to a date) than either the control (60% agreed to a date) or
thunder condition (50% agreed to a date).

The focus on structured courtroom settings and the variability in the effectiveness
of the strategy seen across context potentially restricts the generalizability of the strategy
to these more formal communication constructs. Furthermore, research has demonstrated
variable findings regarding whether or not an increase in perceived source credibility
mediates the link between stealing thunder and persuasion (Dolnik et al., 2003; Williams

et al., 1993). Nevertheless, concluding that the cognitive processes mediating the effect

of stealing thunder on persuasion are variable would be premature given the majority of
the conflicting findings come from a limited number of independent studies in a limited
array of contexts employing different measures of persuasiveness and credibility.
Therefore, this experiment is designed to examine stealing thunder across three contexts
(legal, interpersonal, and relational), while measuring the agent’s perceived
persuasiveness, perceived source credibility, and information valence in a consistent
manner across those contexts.

As mentioned previously, research has demonstrated inconsistency in the degree
to which perceived source credibility mediates the relationship between stealing thunder
and persuasiveness. Perceived source credibility is defined as an individual’s tendency to
make two attributions towards an agent; one, that the agent is a source of a valid message
and two, that the agent intends to deliver those messages they consider most valid
(Hovland, Janis, & Kelley, 1953). In other words, Hovland et al. (1953) suggests that
credible sources are expert and trustworthy. Reinforcing Hovland et al.’s (1953) early
work, O’Keefe (2002) suggests, expertise is an assessment of whether the agent is in a
position to know the truth, to know what is right, or correct, and trustworthiness is an
assessment of whether the communicator will likely be inclined to tell what they see as
the truth. Although others have suggested various alterations to the factor structure of
perceived source credibility (Berlo, Lemert, & Mertz, 1970; McCroskey, 1966;
McCroskey, Hamilton, & Weiner, 1974), all contain features of Hovland et al.'s (1953)
original conceptualization. Therefore, this study conceptualizes perceived source

credibility in the Hovland et al. (1953) tradition.

Being partly interested in the effects of revealed information on source credibility,
Eagly, Wood, and Chaiken (1978) had participants read the transcript of a presentation in
which the agent argued for one of two positions. Prior to reading the transcript the
participants were informed that the background of the agent was either consistent or
inconsistent with the position advocated in the presentation. They were also told that the
audience for the presentation was either for or against the position advocated in the
presentation. In doing so Eagly et al. (1978) created the presence or absence of expected
biases that the agent had for presenting the information in a manner consistent with what
they believed and with what they thought the audience wanted to hear. When biases were
expected but not fulfilled, Eagly et al. (1978) found that participants viewed the agent not
only as more persuasive but also more honest, sincere, open-minded, and consistent than
when the biases were expected and fulfilled. In other words, it could be said one of
Eagly et al.’s (1978) findings was that agents who are perceived to speak against their
own self-interest are seen as more credible.

Consistent with Eagly at al.’s (1978) findings, Cialdini (1993) presents an
example of a waiter he observed who was consistently tipped more than the other waiters
he worked with. The waiter’s strategy was that when the dish a customer ordered was
not particularly good, he would inform the customer of this and instead of recommendin g
a more expensive meal, he was sure to recommend a meal slightly less expensive than the
one the customer originally ordered. This way, instead of appearing as if he wanted to
line his pockets by suggesting more expensive dishes, the waiter appeared to the
customers as if he was making a sacrifice to himself in order to provide the best of what

the restaurant had to offer. After all, waiters are commonly tipped as a percentage of the

total bill. In doing this, Cialdini (1993) concluded the customers made the attribution that
not only was the waiter an authority on the restaurants fare for that evening, but also that
the waiter was acting in the customers best interests. The creation of these attributions,
lead to larger orders and proportionately bigger tips. Taken together, Eagly et al.’s
(1978) and Cialdini’s (1993) findings suggest that agent’s who are perceived as relative
authorities on a topic and who speak against one’s own interests are likely to be
perceived as more credible and hence more persuasive than those who do not.

Extending Eagly et al.’s (1978) argument Williams et al. (1993) hypothesized that
stealing thunder derives its effectiveness in a similar manner. Specifically, they proposed
a model in which stealing thunder leads to increased perceived source credibility which,
in turn, leads to increased persuasion, and tested it in two experiments, one involving a
criminal trial and the second involving a civil suit. Perceived source credibility was
found to mediate the relationship as predicted in study one and in study two. Recently
Dolnik et al. (2003) challenged this view. Specifically, they tested the possibility that
perceived source credibility mediated the relationship using Baron and Kenny’s (1986)
criteria for mediation. The analysis involved demonstrating that (a) stealing thunder,
relative to thunder, has a statistically significant effect on the persuasiveness of the
message, (b) stealing thunder, relative to thunder, has a statistically significant effect on
the perception of source credibility, and (c) when perceptions of source credibility are
controlled, there is a statistically significant decrease in the effect of stealing thunder on
the message’s persuasiveness. When these analyses were conducted by Dolnik et a].
(2003) it was found that perceived source credibility did not meet the third of Barron et

al.’s (1986) criteria.

There are at least two reasons why Dolnik et al.’s (2003) test is inadequate. First,
as conceptualized, source credibility is made up of two dimensions, expertise and
trustworthiness. Thus, if the two dimensions have different effects on persuasiveness and
credibility is not broken down across the two dimensions, it would be possible for the
same credibility score to have different effects. For example, Williams et al. (1993)
combined three single item measures of the agent’s trustworthiness, preparedness, and
convincingness because they were intercorrelated with r’s =.7. In a separate instance,
Dolnik et al. (2003) is ambiguous about how credibility was assessed. It appears that
responses to a set of items were summed because they were intercorrelated and assumed
to measure the same underlying factor. In neither study were indicies of reliability or
evidence of construct validity presented. As a result it is possible that two numerically
identical perceived source credibility scores on either of the two scales could represent
different combinations of expertise and trustworthiness attributions. Additionally, the
perceived source credibility scale in Williams et al. (1993) could be measuring a different
construct than the scale employed in Dolnik et al. (2003). Consequently the possibility
exists that perceived source credibility mediated the relationship between stealing thunder
and persuasion in both Williams et al. (1993) and Dolnik et a1. (2003), but the lack of
reliability, validity, or both allowed them to identify the relationship in only one of the
two studies. Second, Barron et al.’s (1986) third criteria for mediation does not allow for
the fact that multiple mediators exist in the relationship, so that the effect of the first
variable in a causal chain on the last variable in the chain can be statistically significant
when controlling for the effect of a single mediator. Williams et al. (1993) briefly

consider one possible additional mediator in their discussion. They suggest that targets

could also modify the meaning of the information to be consistent with their perception
of the agent. Therefore, to resolve these issues, this study employs a two factor measure
of perceived source credibility based on Hovland et al.’s (1953) conceptualization and
considers a closer look at Williams et al.’s (1993) meaning change idea.

In addition to being placed in a situation where source credibility may be
questioned, targets of a stealing thunder message are placed in a situation in which the
positive impression of agents formed by revelation of negative information that appears
to be against their best interests conflicts with the negative information revealed.
Previous research suggests that people can and do change their perceptions of
information depending on characteristics of the source of the information (Walster,
Aronson, & Abrahams, 1966). One example is the halo effect (Erber & Erber, 2001), in
which lasting impressions of others are affected by first impressions. Asch (1948)
suggests that this effect may occur because the perceived valence of a message can
change by changing the context within which it is expressed. Because speaking against
one’s self-interest causes the agent to be perceived as more honest (Cialdini, 1993;
Williams et al., 2001), credible (Eagly et al., 1978), and likable (Ondrus et a1, 1995)
speaking against one’s self-interest may also produce positive change in valence of the
message presented by creating a positive context in which the message is revealed, and
thereby cause the information to be perceived as less negative.

The idea that the negative information revealed through stealing thunder is
perceived as less negative by its targets than when the information is revealed by an
outside source is consistent with the literature (Dolnik et al., 2003; Williams et al., 1995).

Change of meaning has not, however, been incorporated into a model of stealing thunder.

Given the extant literature, it is proposed the basic act of stealing thunder, that is,
revealing information that apparently is against ones best interests, leads to a more
favorable perception of source credibility, specifically trustworthiness and expertise
(Cialdini, 1993; Eagly et al., 1978). More favorable perceived source credibility results
in less negative views of the information revealed (Asch, 1948; Erber et al., 2001). To
the extent that the information is seen as positive, persuasion follows resulting from the
general positive halo surrounding the agent (Erber et al., 2001; Kelman & Hovland,

1953). Figure 1 presents a visual description of this model.

 

 

 

 

Stealing Source Information Persuasion
Thunder Credibility Valence

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 1. Proposed model: Influence of stealing thunder, source credibility, and
information valence on persuasion.

This study attempts to control systematically the two factors of perceived source
credibility, trustworthiness and expertise, in a manner consistent with past stealing
thunder research. Such an effort, if successful, allows one to partition out their effects.
Hovland et al.’s (1953) conceptualization of trustworthiness implies that an agent should
be perceived as intending to deliver an honest message to the extent that the target
perceives there would be no reason to do otherwise. If the target judges that the agent is
delivering a message in an attempt to mislead or provide false information, the target will
perceive the agent to be low in trustworthiness. O’Keefe (2002) points out that agents for
whom the targets have no expectations are perceived to be the most trustworthy.
Therefore, this study controls trustworthiness by varying whether or not the target was
aware of the fact that negative information about the agent could be potentially be

revealed by somebody else. This reasoning suggests that to the extent that the target is

aware and expecting that the information can be revealed by somebody else, the target
will conclude that the agent is revealing the information for their own personal gain and
thus be perceived as less trustworthy.

In some previous investigations of stealing thunder the target was aware, at least
implicitly, that the information was likely to be revealed by another (Dolnik et al., 2003;
Williams et al., 1993). Furthermore, in other studies participants were apparently
unaware the negative information could be revealed by another (Ondrus et al., 1995;
Zablocki, 1996). The effectiveness of stealing thunder in the studies in which the target
was aware that the information could potentially be revealed by an outside source is
typically less than those studies in which the target was naive to the possibility that the
information could be revealed. Thus, a main effect for the awareness induction is
expected.

The second dimension of perceived source credibility, expertise, is controlled by
varying whether or not the agent offers a refutation for the negative information revealed
in the stealing thunder message. Expertise is considered to be the ability of an agent to
communicate what is correct, or the ability to be in a position to know what is correct
(Hovland et al., 1953). By offering a refutation, that is to deny the accuracy of the
information by providing further evidence (McGuire, 1964), agents assert they have
additional information and implicitly suggest they are in a better position to know what is
correct, both of which should lead to more positive perceptions of expertise. Previous
research has employed a method similar to providing a refutation. In one experimental
condition in Dolnik et al. (2003) the agent put a positive spin on negative information and

framed stealing thunder information about the agent drinking before driving by having

10

the agent state, “I had a couple of drinks at the bar, but not enough to become intoxicated.
Therefore, I felt it was safe to drive.” The authors reported an increase in perceived
source credibility in this condition. Although drawing valid conclusions from this
finding alone is tenuous given the measurement of perceived source credibility, the
findings are intriguing in that stealing thunder studies including a refutation cOndition
report the strategy more effective than when used without framing (Dolnik et al., 2003).
Consequently, the existence of a main effect for the refutation induction is hypothesized.
Furthermore, it is hypothesized that awareness and refutation will have additive
effects on source credibility. The additive main effects of awareness and refutation will
result in high perceived source credibility when the awareness and refutation inductions
are present, low perceived source credibility when the awareness and refutation A
inductions are absent, and moderate perceived source credibility when one or the other

inductions. Table 1 presents the hypothesized relationship.

Table l

Hypothesized Additive Relationship between T rustworthiness and Expertise Inductions
and Perceived Source Credibility

 

 

 

Trustworthiness
Expertise Na‘r've Aware
Refutation High Moderate
No Refutation Moderate Low

 

Note. Levels reported in cells represent hypothesized level of perceived source credibility.

ll

Given the model of stealing thunder specified earlier in this study, it is also
hypothesized that the two main effects for refutation and awareness correspond with the
levels of valence change as presented in Table 2 and perceived persuasiveness for each

condition as presented in Table 3.

Table 2

Hypothesized Additive Relationship between T rustworthiness and Expertise Inductions
and Information Valence Change

 

 

 

Trustworthiness
Expertise Naive Aware
Refutation High Moderate
No Refutation Moderate Low

 

Note. Levels reported in cells represent hypothesized level of information valence change.

Table 3

Hypothesized Additive Relationship between T rustworthiness and Expertise Inductions
Resulting Perceived Persuasiveness

 

 

 

Trustworthiness
Expertise Na’r‘ve Aware
Refutation High Moderate
No Refutation Moderate Low

 

Note. Levels reported in cells represent hypothesized level of perceived persuasiveness.

12

METHOD

Design

The study employs an awareness (naive, aware) X refutation (no refute, refute) X
context (legal, interpersonal, relational) mixed groups design with two offset controls.
The awareness and refutation inductions are independent groups factors and the context
factor is a repeated measure. When crossed, the two levels of awareness and the two
levels of refutation form independent groups. For each context there are two offset
control conditions (no stealing thunder and thunder).
Participants

A convenience sample of 180 volunteer undergraduate students at a large
Midwestern university were assigned randomly to one of the six sets of conditions with
the constraint that all conditions contain an equal number of participants (P5). P5 were
given course credit to compensate them for their time. The sample consisted of 53 males
and 127 females. P3 in the sample were on average 21.33 years old (SD = 2.75) and had
attended college for 3.41 years (SD = 1.18). Furthermore, all but three Ps originated
from the continental United States.
Procedure

The study was conducted in the students’ classrooms. Ps agreed to participate in
a study described as an investigation of the social perceptions of others. After greeting
the Ps and soliciting their participation the experimenter (E) described the required tasks.

Next, every desk in the classroom was arranged to face the front of the room so as
to limit interaction among the Ps. The E also directed the Ps to place everything under

their desks. Subsequently, the E handed out pencils and consent forms. After reviewing

l3

the consent form the E collected them and administered the study materials. The study
materials for each of the six conditions were administered in a packet. To randomly
assign Ps to a condition, each packet was randomly assigned a number from 1 through
180. Packets were then administered in order of their number assignment.

The packet contained the three different interaction scenarios, each containing the
assigned awareness X refutation inductions, or if the case, either of the two control
conditions. Measures of perceived source credibility, perceived valence of the stealing
thunder information and the agent’s perceived persuasiveness separated each of the three
vignettes. Finally, the induction checks and general demographic questions completed
the packet. To control for order effects, the arrangement of the three vignettes in the
packet was counterbalanced.

After reviewing the instructions the E solicited for any final questions.
Immediately before directing them to begin, the E instructed them to remain seated until
they finished their packet. At that time they handed in their packet and returned to their
seat. The E subsequently took an unobtrusive position in the front of the room while the
Ps worked through the study materials. When every P had finished, the E distributed
debriefing forms. Upon dispensing the forms the E reviewed them with the Ps before
fielding questions.

Interaction Scenarios

Each context (see Appendices A, B, and C) centered around an interaction
scenario between two people, e. g., a manager and a potential hire, and defined the P as a
third party. The format of the three contexts, exemplified by the control condition

lacking the thunder information, involved the discussion of a topic or the argument of a

14

point that the P was asked to settle. Although each interaction context involved different
people and situations, length and complexity of the structure remained similar across
contexts.

There were six conditions for each context, the two controls and four stealing
thunder conditions. The six conditions were the control condition-no information, the
control condition-thunder, the na'r've/unrefuted stolen thunder, the naive/refuted stolen
thunder, the aware/unrefuted stolen thunder, and the aware/refitted stolen thunder. The
six conditions for a given context differ only in inclusion or exclusion of one to three
sentences. For example, the refutation condition built upon the no refutation condition by
taking the stolen thunder in the no refute condition and tagging a refutation onto it.
Otherwise, the wording of the inductions retained as much consistency and similarity
between conditions as possible. Nothing else changed between conditions for each
interaction context.

Perceived Persuasiveness Instrument

After reading each scenario, Ps responded to an item assessing the perceived
persuasiveness of the agent (Appendix D). Although thunder was not stolen in the no
thunder control condition, Ps provided a judgment of perceived persuasiveness of the
same agent who stole thunder in the other conditions of that context. The item requested
that the P rate the agent’s persuasiveness on a scale of 1 to 100, where 1 is the least
persuasive and 100 is the most persuasive. This format was employed to replicate the

successful use of similar items in previous work on the topic (Dolnik et al., 2003).

15

Source Credibility Instrument

The source credibility measure was a 20-item instrument (Appendix E) comprised
of 10 expertise items and 10 trustworthiness items. Ps responded on a seven-point Likert
response scale.
Information Evaluation Instrument

Ps also reported their evaluation of the valence of the stolen thunder. This
measure assessed how positively or negatively Ps perceived the stolen thunder. A six-
item Osgood semantic differential scale (Appendix F) with accompanying seven-point
response scales was employed.
Induction Check Instrument

Ps also completed a short induction check consisting of six items (Appendix G).
Item one asked whether or not the P was able to recall the stolen thunder. Items two and
three checked on the awareness induction. Items four and five examined the refutation
induction, and the last item measured the mundane realism of the scenarios. Ps
completed the induction check items fi'om memory without referring to the interaction
scenario they just read.
Demographics Instrument

The final instrument was the demographics measures (Appendix H). There were
four standard demographic questions including age, sex, years in college, and geographic

origin.

16

RESULTS

Source Credibility Measurement Model

Confirrnatory factor analysis was employed to test the content validity of the
source credibility measure. Initially, the two factor source credibility model was
examined separately in each of the three contexts. It was found that deleting expertise
items 2 and 9 as well as trustworthiness items 1, 2, 9, and 10 produced an acceptable two
factor solution (legal: RMSE = .09; interpersonal: RMSE = .08; relational: RMSE = .08).
Nevertheless, observing the correlation matrix, and noting the substantial correlation
between the two factors (legal: r = .94, r' = 1.13; interpersonal: r = .88, r’ = 1.08;
relational: r = .99, r' = 1.20), suggested that either a one factor or a second order
unidimensional solution was feasible. Therefore, the simplest model, the one factor
model was also tested. Once again after the deletion of six items the data were consistent
with the model (legal: RMSE = .08; interpersonal: RMSE = .07; relational: RMSE = .07).
Factor loadings are presented in Table 4. Hence the source credibility measure was
treated as single dimension and the fourteen items were summed to create an index. This
index was distributed normally in each of the three contexts, and had means and standard
deviations of: M = 3.85, SD = .88 (legal), M = 3.93, SD = .86 (interpersonal), and M =
4.85, SD = .78 (relational). Reliability was estimated by coefficient alpha and found to
be: or = .90 (legal), or = .88 (interpersonal), and a = .90 (relational).
Information Evaluation Measurement Model

Confinnatory factor analysis was also employed to test the content validity of the
information evaluation measure. The one factor information evaluation model was

examined separately for each context. Deleting item 3 produced an acceptable one factor

17

Table 4

Factor Loadings for the Unidimensional Source Credibility Solution by Context

 

 

Item number Legal Interpersonal Relational
Expertise #1 .56 .51 .55
Expertise #3 .52 .45 .69
Expertise #4 .75 .65 .77
Expertise #5 .79 .72 .76
Expertise #6 .46 .55 .47
Expertise #7 .42 .42 .35
Expertise #8 .58 .55 .64
Expertise #10 .67 .44 .57
Trustworthiness #3 .60 .68 .55
Trustworthiness #4 .71 .68 .69
Trustworthiness #5 .66 .63 .72
Trustworthiness #6 .66 .75 .73
Trustworthiness #7 .69 .57 .69
Trustworthiness #8 .78 .71 .70

 

Note. Expertise items 2 and 9 along with trustworthiness items 9 and 10 were removed because of
consistently weak factor loadings in each of the three contexts. Trustworthiness items 1 and 2 were
removed because those items had a tendency to produce large errors. Reviewing the face validity of these
six items reveals they are all rather ambiguous in terms of what they were asking respondents to report on.

solution (legal: RMSE = .04; interpersonal: RMSE = .03; relational: RMSE = .02). Factor
loadings are presented in Table 5. The remaining five items were summed to create an

index. The index was distributed normally in all three contexts, and had means and

18

standard deviations of: M = 3.15, SD = 1.46 (legal), M = 3.22, SD = 1.71 (interpersonal),
and M = 3.44, SD = 1.43 (relational). Reliability was estimated by coefficient alpha and

found to be: a = .87 (legal), ()1 = .90 (interpersonal), oz = .87 (relational).

Table 5

Factor Loadings for the Unidimensional Information Evaluation Solution by Context

 

 

Item number Legal Interpersonal Relational
Valence #1 .72 .74 .66
Valence #2 .69 .75 .71
Valence #4 .79 .92 .78
Valence #5 .84 .91 .92
Valence #6 .71 . .71 .71

 

Note. Valence item 3 was removed because the item had a tendency to produce large errors. Reviewing
the face validity of item 3 supports this decision because the item is ambiguous in terms of what it asks
respondents to report on.

Induction Checks

The Ps ability to recall presentation of the negative information across conditions
was tested in each of the three contexts by examining responses to the first induction
check item. In the legal context 84% of those presented with the negative information
recalled the information in their response to the induction check, only .03% of the Ps in
the no information control recalled negative information. Furthermore, a chi-square
analysis performed on these legal context data revealed a statistically insignificant

difference in ability to recall the negative information across the four awareness by

19

refutation conditions and thunder condition, )f (4, N = 150) = 3.09, ns. Taken together,
these analyses suggest the data are consistent with the claim that presentation of the
negative information was effective in the legal context. Performing the same analyses on
the interpersonal data revealed 98% of those presented with the negative information
recalled said information and only .03% of the Ps in the no information control recalled
negative information. A chi-square analysis demonstrated a statistically insignificant
difference in ability to recall the negative information across the four awareness by
refutation conditions and thunder condition, )8 (4, N = 150) = 2.05, ns. Again, these
analyses suggest the data are consistent with the claim that presentation of the negative
information was also effective in the interpersonal context. Similar results were found
for the relational context data where 98% of those presented with the negative
information recalled it on the induction check and none of the Ps in the no information
control recalled negative information. A chi-square analysis performed on the relational
context data revealed a statistically insignificant difference in ability to recall the negative
information across the four awareness by refutation conditions and thunder condition, x2
(4, N = 150) = 2.05, ns. All together, these analyses suggest the data are consistent with
the claim that presentation of the negative information was effective in the relational
context.

The validity of the awareness induction was tested in each of the three contexts by
examining its effects on the sum of responses to the two awareness induction check
items. A two-way analysis of variance performed on the legal context data produced a
main effect for the awareness induction, F (1, 116) = 4.78, p < .05, n2 = .04, a statistically

insignificant effect of the refutation induction, F (1, 116) = 1.19, ns, 172 = .01, and no

20

evidence of an awareness X refutation interaction, F (1, 116) = 0, ns, 112 = 0. Thus, the
data are consistent with the claim that the awareness induction was effective in the legal
context. Considering the interpersonal context data, a two-way analysis of variance
performed also produced a main effect for the awareness induction, F (l , 116) = 5.87, p <
.05, n2 = .05, a statistically insignificant effect of the refutation induction, F (1, 116) =
.80, ns, 172 = .006, and no evidence of an awareness X refutation interaction, F (1, 116) =
.80, ns, 172 = .006. Thus, these data are also consistent with the claim that the awareness
induction was effective in the interpersonal context. Finally, a two—way analysis of
variance performed on the relational context data produced a main effect for the
awareness induction, F (1, 116) = 24.32, p < .05, 112 = .17, a statistically insignificant
effect of the refutation induction, F (1, 116) = .27, ns, 772 = .002, and no evidence of an
awareness X refutation interaction, F (1, 116) = .61, ns, n2 = .004. Hence, the data are
consistent with the claim that the awareness induction was effective in the relational
context.

The validity of the refutation induction was tested in each of the three contexts by
examining its effects on the sum of responses to the two refutation induction check items.
A two-way analysis of variance performed on the legal context data produced a main
effect for the refutation induction, F (1, 116) = 46.63, p < .05, 112 = .29, a statistically
insignificant effect of the awareness induction, F (1, 116) = .42, ns, 772 = .003, and no
evidence of a refutation X awareness interaction, F (1, 116) = .08, ns, 172 = .001. Thus,
the data are consistent with the claim that the refutation induction was highly effective in
the legal context. Performing a two-way analysis of variance on the interpersonal context

data also produced a main effect for the refutation induction, F (1, 116) = 250.73, p < .05,

21

n2 = .68, a statistically insignificant effect of the awareness induction, F (l, 116) = 2.84,
ns, 112 = .008, and no evidence of a refutation X awareness interaction, F (l, 116) = .45,
ns, 112 = .001, revealing that the data are consistent with the claim that the refutation
induction was highly effective in the interpersonal context. A two-way analysis of
variance performed on the relational context data produced a main effect for the
refutation induction, F (l, 116) = 13.23, p < .05, n2 = .10, a statistically insignificant
effect of the awareness induction, F (l, 116) = 1.28, ns, 172 = .01, and no evidence of a
refutation X awareness interaction, F (1, 116) = .39, ns, 172 = .003. Hence, the data are
consistent with the claim that the refutation induction was effective in the relational
context.

The ease with which Ps were able to imagine themselves in the role asked of them
for each condition was tested by examining responses to the final induction check item
separately for each context. A one-way analysis of variance conducted on the legal
context data produced a statistically significant effect for condition, F (5, 174) = 2.63, p <
.05, n2 = .07. A follow up two-way analysis of variance revealed a statistically
significant refutation induction by awareness induction interaction, F (l, 116) = 12.39, p
< .05, n2 = .10. Hence, the data are not consistent with the claim that case of presence
across conditions was equivalent. Even so, Ps were able to imagine themselves in the
situation (M = 4.33, SD = 1.42). A one-way analysis of variance conducted on the
interpersonal context data produced a statistically insignificant effect for condition, F (5,
174) = 1.26, ns, 712 = .03. Thus, the data in the interpersonal context are consistent with
the claim that case of presence across condition was equivalent. Additionally, Ps found it

fairly easy to imagine themselves in the situation (M = 4.50, SD = 1.74). A one—way

22

analysis of variance conducted on the relational context data produced a statistically
insignificant effect for condition, F (5, 174) = .84, ns, 112 = .02. Therefore, the data in the
relational context are also consistent with the claim that ease of presence across condition
was equivalent. Furthermore, Ps found it fairly easy to imagine themselves in the
situation (M = 5.53, SD = .90).
Hypothesis Tests

The hypotheses, that awareness yields lower credibility judgments than
unawareness, refutation produces higher credibility judgments than no refutation, and the
lack of a reason to expect an interaction were tested in each of the three contexts by
examining the awareness and refutation inductions effects on the P’s reports of source
credibility. A two-way analysis of variance performed on the legal context data produced
a statistically insignificant effect for the refutation induction, F (l, 116) = .66, ns, 112 =
.006, a statistically insignificant effect of the awareness induction, F (l, 116) = .45, ns, 112
= .004, and no evidence of a refutation X awareness interaction, F (l, 116) = .02, ns, 172 =
.0001. Thus, the data are not consistent with the claim that refutation and naivety
produce higher credibility judgments in the legal context. The same analysis was
performed on the interpersonal context data which resulted in a statistically insignificant
effect for the refutation induction, F (l, 116) = .49, ns, 172 = .004, a statistically
insignificant effect of the awareness induction, F (1, 116) = 1.70, ns, 112 = .01, and no
evidence of a refutation X awareness interaction, F (l, 116) = .32, ns, 172 = .004.
Consequently, these data are also not consistent with the claim that refutation and naivety
produce higher credibility judgments in the interpersonal context. Performance of a two-

way analysis of variance on the relational context data produced a statistically

23

insignificant effect for the refutation induction, F (1, 116) = .54, ns, 112 = .004, a
statistically insignificant effect of the awareness induction, F (1, 116) = .56, ns, 172 = .004,
and no evidence of a refutation X awareness interaction, F (l, 116) = 1.73, ns, 772 = .01.
The data do not support the claim that refutation and naivety produce higher credibility
judgments in the relational context. Table 6 presents means and standard deviations.

The hypotheses, that awareness yields more negative information valence
judgments than unawareness, refutation produces more positive information valence
judgments than no refutation, and the absence of basis to expect an interaction were
tested in each of the three contexts by examining the awareness and refutation inductions
effects on the P ’5 reports of information valence. A two-way analysis of variance
performed on the legal context data produced a statistically insignificant effect for the
refutation induction, F (1 , 116) = .19, ns, 172 = .002, a statistically insignificant effect of
the awareness induction, F (1, 116) = .61 , ns, 172 = .005, and no evidence of a refutation X
awareness interaction, F (l, 116) = 3.08, ns, 112 = .03. Thus, these data are not consistent
with the claim that refutation and naivety produce more positive valence judgments in the
legal context. A two-way analysis of variance performed on the interpersonal context
data also produced statistically insignificant effects for the refutation induction, F (1,

116) = .07, ns, 172 = .0006, the awareness induction, F (l, 116) = 2.86, ns, 172 = .02, and
the refutation X awareness interaction, F (1, 116) = .001, ns, 112 = .000004. Accordingly,
the data are not consistent with the claim that refutation and naivety produce more
positive valence judgments in the interpersonal context either. Performance of a two-way
analysis of variance on the relational context data produced a statistically insignificant

effect for the refutation induction, F (1, 116) = 2.25, ns, 172 = .02, a statistically

24

Table 6

Observed Relationship between Experimental Inductions and Perceived Source

Credibility

 

 

Trustworthiness Offset Controls
Expertise Na'r've Aware No Information Thunder
Refutation M: 4.00 M: 3.87 M = 3.90 M= 3.71

SD= 1.05 SD= .83 SD= .85 SD= .85
No Refutation M: 3.85 M= 3.76

SD = .97 SD = .76

 

Note. Legal Context. Data reported in cells represent observed level of perceived source credibility.

 

 

 

Trustworthiness Offset Controls
Expertise Naive Aware No Information Thunder
Refutation M = 4.06 M = 3.78 M = 4.66 M = 3.44

SD = .98 SD = .63 SD = .84 SD = .76
No Refutation M= 3.86 M = 3.78

SD = .83 SD = .62

 

Note. Interpersonal Context. Data reported in cells represent observed level of perceived source
credibility.

25

Table 6 (cont’d).

 

 

 

Trustworthiness Offset Controls
Expertise Naive Aware No Information Thunder
Refutation M: 5.04 M: 4.75 M= 4.76 M= 4.56

SD = .78 SD = .82 SD = .70 SD = .79
No Refutation M = 4.96 M = 5.04

SD = .74 SD = .82

 

Note. Relational Context. Data reported in cells represent observed level of perceived source credibility.

insignificant effect of the awareness induction, F (1, 116) = .35, ns, 712 = .003, and no
evidence of a refutation X awareness interaction, F (1, 116) = .53, ns, 112 = .004. Hence,
the data are not consistent with the claim that refutation and naivety produce more
positive valence judgments in the relational context. Means and standard deviations are
presented in Table 7.

The hypotheses, that awareness yields less perceived persuasiveness than
unawareness, refutation produces more perceived persuasiveness than no refutation, and
the lack of grounds to expect an interaction were tested in each of the three contexts by
examining the awareness and refutation inductions effects on the P ’3 reports of perceived
persuasiveness. A two-way analysis of variance performed on the legal context data
produced a statistically insignificant effect for the refutation induction, F (1, 116) = .56,
ns, 172 = .005, a statistically insignificant effect of the awareness induction, F (1, 116) =

.77, ns, 172 = .006, and no evidence of a refutation X awareness interaction, F (l , 116) =

26

Table 7

Observed Relationship between Experimental Inductions and Information Valence
Judgment

 

 

Trustworthiness Offset Controls
Expertise Naive Aware No Information Thunder
Refutation M= 2.65 M = 3.28 M = N/A M = 3.43

SD= 1.35 SD= 1.49 SD=N/A SD= 1.64
No Refutation M = 2.98 M = 2.74

SD= 1.36 SD= 1.20

 

Note. Legal Context. Data reported in cells represent observed level of information valence.

 

 

 

Trustworthiness Offset Controls
Expertise Naive Aware No Information Thunder
Refutation M= 2.99 M= 2.55 M= N/A M: 3.03

SD = 1.46 SD = 1.25 SD = N/A SD = 1.54
No Refutation M = 2.91 M = 2.49

SD= 1.52 SD= 1.29

 

Note. Interpersonal Context. Data reported in cells represent observed level of information valence.

27

Table 7 (cont’d).

 

 

 

Trustworthiness Offset Controls
Expertise Naive Aware No Information Thunder
Refutation M= 3.03 M: 3.06 M= N/A M: 3.37

SD=1.51 SD=1.24 SD=N/A SD=1.40
No Refutation M = 3.60 M = 3.26

SD=1.52 SD=1.35

 

Note. Relational Context. Data reported in cells represent observed level of information valence.

.22, ns, 112 = .002. Thus, the data are not consistent with the claim that refutation and
naivety produce more perceived persuasiveness in the legal context. A two-way analysis
of variance performed on the interpersonal context data also produced a statistically
insignificant effect for the refutation induction, F (1, 116) = .60, ns, 112 = .005, a
statistically insignificant effect of the awareness induction, F (1, 116) = .11, ns, 712 =
.0009, and no evidence of a refutation X awareness interaction, F (l, 116) = .32, ns, 172 =
.003. Therefore, the data are not consistent with the claim that refutation and naivety
produce more perceived persuasiveness in the interpersonal context either. A two-way
analysis of variance performed on the relational context data produced a main effect for
the refutation induction, F (l , 116) = 3.97, p < .05, n2 = .03, a statistically insignificant
effect of the awareness induction, F (l, 116) = .33, ns, 172 = .003, and no evidence of a
refutation X awareness interaction, F (1, 116) = .34, ns, 772 = .003. Hence, the data are
not consistent with‘ the claim that refutation and naivety produce more perceived

persuasiveness in the relational context. Table 8 presents means and standard deviations.

28

Table 8

Observed Relationship between Experimental Inductions and Perceived Persuasiveness

 

 

Trustworthiness Offset Controls
Expertise Naive Aware No Information Thunder
Refutation M = 43.40 M = 49.17 M = 47.83 M = 46.37

SD=25.13 SD=21.23 SD=24.55 SD=22.14
No Refutation M = 42.17 M = 43.93

SD = 25.74 SD = 21.93

 

Note. Legal Context. Data reported in cells represent observed level of perceived persuasiveness.

 

 

 

Trustworthiness Offset Controls
Expertise Naive Aware No Information Thunder
Refutation M = 43.93 M = 47.90 M = 66.73 M = 53.23

SD = 24.66 SD = 22.16 SD = 18.08 SD = 30.05
No Refutation M = 43.00 M = 41.93

SD = 26.86 SD = 23.72

 

Note. Interpersonal Context. Data reported in cells represent observed level of perceived persuasiveness.

29

Table 8 (cont’d).

 

 

 

Trustworthiness Offset Controls
Expertise Naive Aware No Information Thunder
Refutation M: 56.10 M= 51.20 M: 58.93 M= 50.30
SD = 21.38 SD = 22.84 SD = 22.47 SD = 23.00
No Refutation M = 62.10 M = 62.13

SD = 26.89 SD = 21.85

 

Note. Relational Context. Data reported in cells represent observed level of perceived persuasiveness.

In order to make claims regarding the effects of stealing thunder on each of the
three dependent variables, each of the control conditions were compared against all other
conditions in that context using Dunnett’s t in each context. Analyses performed on the
data in the legal context revealed in the case of each of the dependent variables,
statistically insignificant differences between both control condition means and each of
the other condition means. Analyses on the interpersonal context data revealed
insignificant differences between the thunder control condition and each of the
experimental condition means for source credibility with the exception of the difference
between the naive by refutation condition and the thunder control condition means, d0 =
|.49|, do = .62, p < .05. On the other hand, differences between the no information
control condition and each of the experimental condition means for source credibility
were significant. Table 9 reports the Dunnett’s t values. Examination of the information

valence variable revealed insignificant differences between the control conditions and

30

Table 9

Dunnett 's T Results Comparing Source Credibility Mean Difl'erences between the No
Information Control and Other Groups in the Interpersonal Context

 

 

Condition do do Significance

Thunder Control |.50| -1 .21 p < .05

Naive X No Refutation |.50| -.80 p < .05

Aware X No Refutation |.50| -.88 p < .05

Naive X Refutation |.50| -.60 p < .05

Aware X Refutation |.50| -.89 p < .05
Table 10

Dunnett ’s T Results Comparing Perceived Persuasiveness Mean Difi’erences between the
No Information Control and Other Groups in the Interpersonal Context

 

 

Condition do do Significance
Naive X No Refutation |15.46| -23.73 p < .05
Aware X No Refutation |15.46| -24.80 p < .05
Naive X Refutation |15.46| -22.80 p < .05
Aware X Refutation |15.46| -l8.83 p < .05

 

each of the other condition means. Consideration of perceived persuasiveness exposed
insignificant differences between the thunder control condition and each of the other

conditions means. However, statistically significant differences between the no

31

information control condition and each experimental condition mean were found. Table
10 reports the Dunnett’s t values. Analyses on the relational context data uncovered
statistically insignificant differences in the case of each of the dependent variables,

between both control conditions and each of the other conditions means.

32

DISSCUSSION

Primarily, the findings of this study question the robustness of the stealing
thunder effect as reported by Williams et al. (2001). The typical research on the issue
suggests that the revelation of a relevant piece of negative information about oneself will
solicit a persuasive effect. The findings of this study suggest there may be certain
conditions that must be met in order to elicit an effect. Contrary to expectations, the
results of this study found stealing thunder made no sizeable impact on the agent’s
perceived source credibility, the valence of the presented information, or the agent’s
perceived persuasiveness when in conjunction with the presentation of a refutation or
whether the target is naive or aware of the existence of negative information.
Furthermore, the same lack of effect was found when both of the control conditions are
compared against all other conditions. Only two exceptions were found. The first was in
the relational context where the agent was perceived to be significantly more persuasive
when a refutation was not delivered as part of the stealing thunder message relative to
when one was delivered. The second was in the interpersonal context where the agent’s
perceived source credibility and perceived persuasiveness were judged significantly
lower when thunder was stolen than when no information was revealed.

Several features in the design of this study could account for these findings. One
possible explanation for the lack of a stealing thunder effect may be linked to the method
of measurement used to assess the agent’s perceived persuasiveness. Ps were instructed
to choose a number between 1 and 100 to represent how persuasive they felt the agent
was. This large frame of reference could have resulted in different Ps assigning different

meanings to the same number. If this were the case one would expect to observe

33

particularly large standard deviations and a lack of statistical power increasing the
possibility of Type 11 error. Even though large standard deviations averaging
approximately 24 units were observed in support of this line of thought, this does not
seem to be an adequate explanation. Effects of stealing thunder on source credibility and
information valence were also predicted and no effect was measured, even though the
testing of the measurement models for both scales reported psychometric qualities
superior to those of the persuasiveness instrument suggesting the problem lie elsewhere
than the measurement.

Other possibilities are that either the construct validity of the scales used to
measure the dependent variables was questionable resulting in invalid measurement of
the intended concept or that the inductions were simply weak. Either of these situations
could have resulted not finding a hypothesized effect. The evidence in the data suggests
the inductions in this study did not have their intended effects rather than the alternative
explanation of poor measurement. Support for this position comes from examining the
correlations observed between scores on the perceived source credibility scale and the
information valence scale in each of the three contexts (legal: r = .25, r’ = .28;
interpersonal: r = .49, r' = .55; relational: r = .28, r' = .32) as well as those between scores
on the perceived source credibility scale and perceive persuasiveness ratings (legal: r =
.40, r' = .42; interpersonal: r = .42, r’ = .45; relational: r = .26, r' = .27). These
correlations are not only sizeable but also in the direction that would be predicted by
theory.

Additionally, the thunder control condition and the no information control

condition had similar impacts on the dependent variables. One would at the very least

34

 

expect to see a negative impact on the P ’s perceptions of the agent in the thunder control
condition where an outside source presents incriminating information about the agent,
however, that was not the case. Furthermore, although the induction checks suggest that
Ps were accurate in the recall of the various inductions they had or had not been
presented, the effect sizes for several of the conditions were rather small. Therefore, it
appears that although Ps were typically able to recall having seen the inductions, for
some reason the inductions had little effect on the Ps evaluations of the agent and the
information revealed.

It is feasible the inductions failed to have an effect on perceptions of the agent
because stealing thunder effect may not rest in automatic cognitive processes. Studies
conducted to investigate the cognitive processes resulting in the stealing thunder effect
have reported conflicting results. Ondrus et al. (1995) found that when Ps were not
cognitively busy, stealing thunder had a larger effect than when Ps were cognitively
busy. Alternatively, Williams et al. (2001) reports that need for cognition is an unreliable
predictor of stealing thunder’s effectiveness. The conflicting findings could result if
there is an unaccounted for intervening variable impacting the cognitive processes
involved. The findings of this study in relation to others suggests that the intervening
variable may be the target’s level of motivation to process the information presented
which in turn impacts the extent to which targets process the information presented in the
situation.

Consider the experimental materials Ps received in this study, consisting of a 25
page instrument. Based on the E’s observations of some P 's reactions, the sheer amount

of information contained in the instrument could have reduced the motivation of

35

participants to be particularly involved in the task. When verbalized by Ps, reactions to
the instrument ranged from disillusionment to frustration due to the length. F urtherrnore,
33 of the 180 instruments administered had to be recollected due to substantial numbers
of items lefi incomplete. While completion of the questionnaire did not take longer to
complete (25 minutes) than was advertised (30 minutes) Ps at times would note their
frustration with how long it would take to complete the instrument. There was no attempt
by the E at that time to alter the method of data collection because taken individually the
comments did not suggest a problem. Taken as a whole, however, the comments appear
to be symptomatic of the Ps general lack of motivation to be actively involved in
completion of the experimental tasks. Given that the presentation of scenarios was
counterbalanced, the data were reanalyzed post hoc for an order effect only considering
the scenarios that were presented first. No new effects emerged from these analyses and
it appears order of presentation did not play a role. While one would expect an order
effect to emerge if Ps were tiring of the task over the course of completing the instrument
one would not emerge if Ps motivation to complete the task was eradicated upon initial
exposure to the instrument when it was placed upon their desk. Given the evidence this
could have easily been the case.

Contrast this with a typical stealing thunder study (Dolnik et al., 2003) in which
Ps are given the very specific tasks of reading an abridged transcript of a courtroom trial,
reporting the guilt or innocence of the defendant, and response on a scale measuring
source credibility. In that study the Ps work on a single court case transcript and task that
is more focused and not as hypothetical as the task in this study. Although, the greater

amount of material presented in the present study would not have necessarily impacted

36

the ability to recall the presentation of the stealing thunder information because the
information was presented at the conclusion of the vignette and Ps were asked to recall
the information after reading only two pages it could have, however, limited the Ps
processing of the information presented in a systematic manner.

This line of thought is consistent with Petty and Cacioppo’s (1986) elaboration
likelihood model which suggests persOnal relevance, need for cognition, personal
responsibility, issue involvement, etc. are important factors in whether or not individuals
are motivated to process incoming information in a controlled fashion. Additionally,
Schwarz (1995) suggests that in forming impressions, our memories of information from
the situation are searched until one feels confident they have considered enough
information to make a judgment. According to Kunda (1999) the level of needed
confidence to stop the search may be determined by the extent to which people are
motivated by an accuracy goal. Accuracy goals lead people to invest greater effort in the
judgment task and to search more thoroughly for the best possible reasoning strategy
(Kunda, 1999). A target’s evaluation of the agent’s message in a stealing thunder
situation would be governed by accuracy goals particularly when the target feels their
decision might cause the agent to be treated unfairly (Kunda, 1999). In sum, if the target
were unmotivated to process in the situation they would be less concerned with whether
or not the agent would be treated unfairly and hence less motivated to weigh all of the
relevant information presented and rather use a heuristic or peripheral cue to make their
evaluation of the agent’s message, as suggested by either Petty and Cacioppo (1986) or
Schwarz (1995) and Kunda (1999). It seems the likely heuristic targets of stealing

thunder might rely on would be to evaluate based on the most salient information

37

presented. This would take the form of what the majority of the information suggests or
what the most outstanding piece suggests.

The research findings of this study and those of other studies are consistent with
this rationalization. For example, situations using hypothetical scenarios would solicit
the smallest effect for stealing thunder. Ps would likely to be least motivated in this type
of setting because there are few consequences of a poor judgment of the agent described
in the vignette. The only time when one would expect to observe a stealing thunder
effect in a hypothetical situation would be if the E did a good job of motivating Ps for the
experimental task or the thunder information was so relevant to the decision at hand that
it became the most salient piece of information revealed and even then there would be
only an effect between when the negative information was revealed and when it was not
revealed. Exactly such effects were observed in this experiment. In the main, no
differences between any conditions were found except for some observed in the
interpersonal context where a person was applying for a bank teller position and they had
been fired in the past for stealing money.

One might expect the effectiveness of stealing thunder to be somewhat better in
situations where targets or P5 are slightly more motivated. These conditions are typically
found in stealing thunder studies conducted by Williams or one of his colleagues.
Typically, Ps in these studies respond to a single scenario involving either a jury member
scenario in which Ps make judgments about a defendant or a political scenario in which
Ps make judgments about a potential candidate. In the vast majority of these studies
stealing thunder was found to have an effect. Taken together, not only could the topics of

the scenarios in the past studies be interpreted to have more at stake for the agent than

38

this study but they also are less demanding of the Ps allowing them to focus more on the
single, shorter, task. Both of these differences between the past and present studies could
lead the past studies to observe the very replicable effects of stealing thunder and the
observations in the current study to fail replicate the effect.

Finally, based on the motivation rationale, in situations involving high motivation
one would expect to observe a large discrepancy between the effectiveness of stealing
thunder and thunder with the effectiveness of the simple persuasive request in the middle.
This pattern of results was exactly what was found in Zablocki (1996). The situation
created this study undoubtedly had the highest levels of motivation and involvement for
P5 primarily because data collection in the Zablocki (1996) study involved face to face
interactions between a confederate and P the stakes for making accurate evaluations of
the agent could have been particularly high. The stakes for making accurate evaluations
were particularly great because evaluating the agent based on a heuristic or peripheral cue
would likely lead to less accurate judgments and taking action based on inaccurate
judgments directly to the agent’s face instead of reporting promised action on a
questionnaire could be perceived as particularly damaging or high risk to the target.

Findings observed in the relational context in the present study are also consistent
with this rationale. Examination of the agent’s persuasiveness ratings in the traditionally
studied stealing thunder conditions (the no information control, thunder, and the naive X
refiitation cOndition) from the relational context reveals the pattern of means, which
approach significance F (2, 87) = 3.59, p = .06, approximate the pattern found by
Zablocki (1996) where stealing thunder is most persuasive, followed by the control

condition, which was subsequently followed by the thunder condition. Even though Ps

39

motivation in this study may have been reduced, the motivation of Ps to evaluate the
vignettes with an accuracy goal or in a controlled manner by weighing all of the
information should have been greatest in the relational context and lowest in the other
two. Kunda (1999) suggests the reason for this is because accuracy goals can arise also
when judgments are made have personal relevance. Given the relational context in this
study involved a dating situation which Ps may have been much more familiar with than
the other two scenarios it is plausible that there was greater personal relevance in the
relational context than in the other two.

Another interesting finding arising from the relational context data was that the
agent was significantly more persuasive when the stolen thunder was not refisted than
when it was refuted (although given the effect size the statistical significance of the
finding may be due to Type I error). After a secondary look at the refutation message in
the relational context it appeared that the refutation was not exactly a refutation but rather
it made the agent appear to be providing an excuse for the thunder information. As a
result, Ps could have paid more attention to the negative information because the weak
refutation made the agent appear as if they were trying to weasel their way out of the
situation. This finding is consistent with some past research where weak refutations were
inadvertently used and no effect for the refutation was found (Williams et al., 2001). A
finding like this is expected here if Ps are basing their judgments on their controlled
processing of the information in the situation rather than on a heuristic.

Considered all together, this study holds some interesting implications for future
work in the area of stealing thunder. Primarily, the findings suggest that motivation to

process the information presented may play a central role in the effectiveness of the

40

strategy. While the development of the initial impressions the target has for the agent
may be developed more or less automatically, the cognitive reevaluation of the
incriminating information revealed via the stolen thunder, and the judgment that is
required to be made based on all of that information may be carried out in a more
controlled fashion. Future research examining this proposition may be quite successful in
producing explanations for some of stealing thunder’s unexpected effects.

One of those unexpected effects found in this study is that some information may
be too negative or incriminating for stealing thunder to be effective. This implies that
some information is better off being not told. This is not necessarily a bad idea
considering some negative information may be considered relevant by some individuals
but in fact be quite the opposite in reality. It would be interesting for future research to
examine exactly how the severity of the negative information plays into the effectiveness
of stealing thunder. The findings of this study also suggest that the perception of a
personal familiarity between the target and agent may increase the effectiveness of
stealing thunder beyond not revealing anything. Based on this information, stealing
thunder’s effectiveness could be increased, regardless, of context if the agent can employ
communication previous to the revelation of negative information that creates the
impression of a personally relevant bond between the target and agent. This in turn
would increase the motivation to process the information presented in the situation by
increasing the extent to which the target has accuracy goals governing their judgment.
Also, this study suggests that weak refutations may exacerbate the effects of the negative

information. Though some of these findings are only trends observed in the data that

41

were consistent with past research, new research specifically investigating them may lead
to interesting results.

Although some of the limitations of this study have been addressed inline with the
discussion there is one overall limitation found across stealing thunder research. There is
a need to investigate stealing thunder behaviorally. There has been a substantial lack of
work taking this perspective. Of approximately 13 stealing thunder studies reviewed,
only one examined the topic by observing compliance with a request employing stealing
thunder. Given the propositions forwarded in the discussion of this paper and the
findings of the Zablocki (1996) study, investigation of stealing thunder in a face to face

communication context may be very interesting.

42

“Iii“!!! tain‘t.“ 1“ "ah. h

APPENDICIES

43

APPENDIX A

Interaction Scenarios -— Legal Context
Control — No Thunder Information

In some instances, legal disputes may be settled through the process of
arbitration. Arbitration commonly is used to settle disputes between two parties
or individuals. There is no jury in arbitration. Instead, an independent individual,
called an arbitrator, listens to arguments from both sides of the dispute and settles
the dispute by making a binding judgment in favor of one side or the other. When
reading the following passage please read as if you were the arbitrator deciding
upon the following matter.

Hays Bicycle Corp., a maker of upper mid-priced bicycles, is at the center
of a class action lawsuit involving customers of theirs who have been injured by
riding a Hays bicycle. The specifics of the injuries are as follows: The injuries
experienced by the customers of Hays result from the pedals on the bicycle
breaking off. Simply, the pedals are one of the parts involved in transferring the
force exerted by the rider’s feet to the chain enabling the bike to travel forward
and as a result they must be strong. Recently the pedals on Hays bicycles have
been breaking while the bicycle is being ridden resulting in injuries, some severe.
For example, some riders have suffered severe groin injuries from falling onto the
top bar of the bicycle due to the pedals breaking away from under them while
riding the bicycle.

The attorney for the injured customers argue that no other bicycle

manufacturer has recently experienced problems like the ones Hays is

44

experiencing with their pedals. They argue that the problems result from Hays
Bicycle Corp.’s slow response to their customer’s complaints and problems.
Furthermore, the attorneys argue that is evidence of the fact Hays does not put the
time and energy needed into producing a quality bicycle, thus resulting in the
injuries being reported. Finally, the attorney for the injured customers argues that
Hayes safety and construction standards are far below what would be expected
from a large bicycle manufacturer.

On the other hand the attorney for Hays Bicycle Corp. has argued that
Hays pedals have never in their history had problems like they are experiencing
now and are working to correct the problem. Additionally, Hays attorneys argue
that all bicycle companies see problems like this with their parts from time to time
and it is not symptomatic of any specific quality problem with Hays Corp.
Furthermore, Hays prides itself on producing a quality product and quality
customer service and it has been slow in responding because of the time needed to
deal with the recent problems with their pedals. Finally, Hays attorneys argue
that the breaking pedals could be a result of customers riding their bikes in
conditions they were not designed for.
For the remaining five conditions, insert text here.

While the remainder of the arbitration case is too extensive to report here,
the previous text is a general summary of the important pieces of the case.
Essentially, the attorney for the customers of Hays Bicycle Corp. would like to

see the arbitrator decide for the customers and award them the monetary

45

settlement. Alternatively, the attorney for Hays Bicycle Corp. would like to see

the arbitrator decide in favor of Hays, relieving them from blame.

Control — Thunder Information

During the process of the arbitration hearing an independent investigation,
not associated with either Hays Bicycle Corp. or the customers of Hays, reveals
that Hays Bicycle Corp. has recently been using pedals that are not as strong as
they have used in the past.
Stealing Thunder —— Naive X No Refute

During the process of the arbitration hearing Hays Bicycle Corp. discloses
that they have recently been using pedals that are not as strong as they have used
in the past.
Stealing Thunder — Aware X No Refute

Both the attorneys for the customers of Hays and the attorneys for Hays
Bicycle Corp. are aware the other has scientific evidence revealing the strength of
the Hays pedals that have been breaking compared with Hays pedals from the past
that have not broken. During the process of the arbitration hearing Hays Bicycle
Corp. discloses that they have recently been using pedals that are not as strong as
they have used in the past.
Stealing Thunder — Nai've X Refute

During the process of the arbitration hearing Hays Bicycle Corp. discloses
that they have recently been using pedals that are not as strong as they have used
in the past and their testing made them confident they would not be so weak as to

lead to breakage.

46

Stealing Thunder — Aware X Refute

Both the attorneys for the customers of Hays and the attorneys for Hays
Bicycle Corp. are aware the other has scientific evidence revealing the strength of
the Hays pedals that have been breaking compared with Hays pedals from the past
that have not broken. During the process of the arbitration hearing Hays Bicycle
Corp. discloses that they have recently been using pedals that are not as strong as
they have used in the past and their testing made them confident they would not

be so weak as to lead to breakage.

47

APPENDIX B

Interaction Scenarios — Interpersonal Context
Control — No Thunder Information
Filling job vacancies and increasing staff size to keep up with demand
become important tasks for any manager of a growing business. Occasionally,
managers seek the advice of a human resource advisor in order to help make a
decision of whether or not to hire an applicant. While you read the following
passage please read as if you are the human resource advisor to the manager and

may be asked for your opinion of whether or not to hire the applicant.

 

The bank manager, Irvine Ewart, of Birrell Bank & Trust, a small bank '3
located in Birrell, Idaho, has for the past 4 months searched for a new bank teller
in order to keep up with the increasing demand the growing city places on the
bank. For Irvine, the relatively small size of the city makes qualified applicants
for the position hard to find.

The application process involves two steps. First a written application,
amounting to a basic personal history, must be submitted to Mr. Ewart for review.
Second, all potential applicants must undergo a personal interview by Mr. Ewart.
Typically, Mr. Ewart finds that applicants rarely make it through the first step of
the application process. On this day, however, Mr. Ewart receives an application
from an individual who appears right for the job. Tony Moyet’s application
appears to be all in order and Mr. Ewart decides to call him to set up a face-to-

face interview.

48

At the interview Mr. Ewart is impressed with Mr. Moyet’s education level.
Mr. Moyet possesses an Associates Degree from a local college in Business
Accounting and because a position as a bank teller involves keeping daily
accounting ledgers of all transactions, Mr. Ewart finds this an important asset for
Mr. Moyet to have considering most applicants thus far barely finished high
school. Furthermore, Mr. Ewart feels that Mr. Moyet’s age, 34, makes the
prospect of hiring Mr. Moyet more attractive. Mr. Ewart feels that his elder status
compared to most other applicants makes him more likely to be level headed and
focused. This is an important consideration when the job being applied for
involves the accurate handling of cash money. Finally, probably the most
important feature that Mr. Moyet would bring to the job if hired pertains to his
previous job experience. Although it not reported where the experience came
from, Mr. Moyet states on his application he previously worked as a bank teller.

Although the interview between Mr. Ewart and Mr. Moyet is
overwhelmingly positive, a couple things make Mr. Ewart slightly less than
impressed with Mr. Moyet. For example, Mr. Moyet was five minutes late for the
interview and he continually mispronounces Mr. Ewart’s name even after hearing
it said correctly by Mr. Ewart’s secretary during the interview. Also, Mr. Moyet
had a slightly disheveled appearance but Mr. Ewart attributes this to Mr. Moyet’s
tardiness.

Just before the interview concludes Mr. Moyet discusses his previous
experiences as a bank teller. He informs Mr. Ewart of his responsibilities and

tasks at his previous job and Mr. Ewart is pleased to learn that he will not need to

49

train Mr. Moyet on many of the responsibilities he would handle if hired given
they are so similar to his previous job.
For the remaining five conditions, insert text here.

The manager, Mr. Ewart, concludes the interview by telling the applicant,
Mr. Moyet, that he is not ready to make a final decision until he has some time to
think about it and receive input from his advisors, but not to worry and that he

will be hearing from them soon.

Control — Thunder Information

During the interview the bank manager, Mr. Ewart, receives a telephone
call revealing that, the interviewee, Mr. Moyet had in the past been fired from the
job similar to the one he is applying for stealing money.
Stealing Thunder — Naive X No Refute

During the interview Mr. Moyet discloses that he had in the past been
fired from the job similar to the one he is applying for stealing money.
Stealing Thunder — Aware X No Refute

Both the bank manager, Mr. Ewart, and the applicant, Mr. Moyet, are
aware that a background check will be performed on all applicants to the bank
teller position Mr. Moyet is applying for. During the interview Mr. Moyet
discloses that he had in the past been fired from the job similar to the one he is

applying for stealing money.

50

Stealing Thunder — Naive X Refute

During the interview Mr. Moyet discloses that he had in the past been
fired from the job similar to the one he is applying for stealing money but that he
was wrongly accused and someone else had stolen the money.
Stealing Thunder —- Aware X Refute

Both the bank manager, Mr. Ewart, and the applicant, Mr. Moyet, are
aware that a background check will be performed on all applicants to the bank
teller position Mr. Moyet is applying for. During the interview Mr. Moyet
discloses that he had in the past been fired from the job similar to the one he is
applying for stealing money but that he was wrongly accused and someone else

had stolen the money.

51

APPENDIX C

Interaction Scenarios — Relational Context
Control — No Thunder Information

From time to time we all seek our friends for advice on our relationships.
Sometimes we are asked for our opinion on what we generally think of our
friend’s significant other while other times see our friends seeking our support for
a difficult decision they may have to make in their relationship. When reading the
following passage, keep that role as a friend in mind as if you were to recommend
to your friend what to do.

A couple weeks ago while out with a large group of friends your fiiend
Chris was introduced to Kelly through a mutual friend. As the night went along
they both seemed to hit it off great and by the end of the night had exchanged
phone numbers and a brief kiss. Chris is fairly open about their relationships with
you and from everything you have heard from Chris seems to suggest the two
have been getting along well since they met. They have been out on a few dates
and have met some of each others fiiends. The two of you both agree that up until
a couple days ago Chris’ budding relationship had been happily progressing as
one would expect, although nothing too serious had happened yet.

Chris consistently raves about Kelly and talks of how fun and entertaining
it is to be together. Chris is always telling you how attractive Kelly is as well.
Not to mention how considerate Kelly seems. One of the most memorable things
Chris has told you so far is how Kelly called the aflemoon after they met to find

Chris was not feeling well. After being invited over, Kelly proceeded to surprise

52

Chris and bring over Chris’s favorite movie so they could relax together while
Chris recovered from the night before. When hearing Chris talk to you about this,
you cannot remember when the last time Chris had been so happy.

On the other hand Chris has also confided in you that they have been
concerned about their new relationship with Kelly. Chris has always felt
somewhat distant from Kelly and is unsure why, almost as though Kelly is
keeping Chris at arms length. So far Chris has attributed this to the short life of
their relationship and assumes it will go away with time. Furthermore, Kelly acts,
although not extreme, in a more controlling way than Chris is used to. Recently,
Chris has noticed that Kelly’s behavior usually istinged with a mild nervousness,
almost as though Kelly is not comfortable.

Chris and Kelly have talked about this and Kelly says that Chris is not the
direct cause of the strange behavior, Chris nonetheless does not find it altogether
pleasing to deal with that. According to Chris, during their talk, Kelly expressed
feelings for Chris and they were nothing less than positive. Kelly said the
problem is simply that they had a lot on their mind.

For the remaining five conditions, insert text here.

Altogether, on one hand Chris seems to be extremely happy in the new
relationship with Kelly, but at the same time is concerned with its fiiture. Chris
being an extremely close fiiend with you has confided all of this in you with the
hope that you will help in the decision of what to do with their budding

relationship with Kelly.

53

Control — Thunder Information

Just alter their talk, Chris overhears a mutual fiiend say that Kelly’s
strange behavior is due to a curable sexually transmitted disease infection Kelly
was recently diagnosed with.
Stealing Thunder — Naive X No Refute

During their talk, Kelly revealed to Chris that the strange behavior is due
to a curable sexually transmitted disease infection Kelly was recently diagnosed
with.
Stealing Thunder — Aware X No Refute

The mutual fiiend that introduced Chris and Kelly to one another knows
why Kelly has been acting distant and both Chris and Kelly are aware of this.
During their talk, Kelly reveals to Chris that the strange behavior is due to a
curable sexually transmitted disease infection Kelly was recently diagnosed with.
Stealing Thunder - Nai’ve X Refute

During their talk, Kelly reveals to Chris that the strange behavior is due to
a curable sexually transmitted disease infection Kelly was recently diagnosed with
but doesn’t see how it is possible and is getting a second opinion.
Stealing Thunder -— Aware X Refute

The mutual friend that introduced Chris and Kelly to one another knows
why Kelly has been acting distant and both Chris and Kelly are aware of this.
During their talk, Kelly reveals to Chris that the strange behavior is due to a
curable sexually transmitted disease infection Kelly was recently diagnosed with

but doesn’t see how it is possible and is getting a second opinion.

54

APPENDIX D

Persuasiveness Instrument
Legal Context
Instructions: On the following item please think about the scenario you just read
and indicate your opinion by writing your response on the blank space provided.
Use a scale of l to 100 where 1 represents not very and 100 represents extremely.
If you have any questions please ask the researcher.

1) How persuasive was the attorney for Hays Bicycle Corp?

 

Interpersonal Context

Instructions: On the following item please think about the scenario you just read
and indicate your opinion by writing your response on the blank space provided.
Use a scale of 1 to 100 where 1 represents not very and 100 represents extremely.
If you have any questions please ask the researcher.

1) How persuasive was Mr. Moyet?

 

Relational Context

Instructions: On the following item please think about the scenario you just read
and indicate your opinion by writing your response on the blank space provided.
Use a scale of 1 to 100 where 1 represents not very and 100 represents extremely.
If you have any questions please ask the researcher.

1) How persuasive was Kim?

 

55

APPENDIX E

Source Credibility Instrument
Instructions: On the items below please think about the scenario you just read and

indicate your feelings about . Please indicate how much you agree or

 

disagree with each statement by circling the response on the scale provided that most

accurately represents how you feel about . If you have any questions

 

please ask the researcher.
Expertise

I) I have confidence in what had to say.

 

Agree Agree Agree Undecided Disagree Disagree Disagree

 

Strongly Somewhat Somewhat Strongly
2) is not very intelligent. (R)
Agree Agree Agree Undecided Disagree Disagree Disagree
Strongly Somewhat Somewhat Strongly
3) seems to be a competent individual.

 

Agree Agree Agree Undecided Disagree Disagree Disagree

 

Strongly Somewhat Somewhat Strongly
4) I feel what said is not reliable. (R)
Agree Agree Agree Undecided Disagree Disagree Disagree
Strongly Somewhat Somewhat Strongly
5) There is little value in what talked about. (R)

 

Agree Agree Agree Undecided Disagree Disagree Disagree
Strongly Somewhat Somewhat Strongly

6) is a well respected individual.

 

Agree Agree Agree Undecided Disagree Disagree Disagree
Strongly Somewhat Somewhat Strongly

56

7) has taken consideration of the factors involved in the situation.

 

Agree Agree Agree Undecided Disagree Disagree Disagree
Strongly Somewhat Somewhat Strongly

8) I feel did not have much skill in getting their point across. (R)

 

Agree Agree Agree Undecided Disagree Disagree Disagree
Strongly Somewhat Somewhat Strongly

9) Not many individuals would be as poised as was in this situation.

 

Agree Agree Agree Undecided Disagree Disagree Disagree
Strongly Somewhat Somewhat Strongly

10) ’s statements carry little clout. (R)

 

Agree Agree Agree Undecided Disagree Disagree Disagree
Strongly Somewhat Somewhat Strongly

T rusworthiness
1) does not appear to be an honest individual. (R)

Agree Agree Agree Undecided Disagree Disagree Disagree
Strongly Somewhat Somewhat Strongly

2) I would be likely to believe what has to say in most circumstances.

 

Agree Agree Agree Undecided Disagree Disagree Disagree
Strongly Somewhat Somewhat Strongly

3) is a person of character.

 

Agree Agree Agree Undecided Disagree Disagree Disagree
Strongly Somewhat Somewhat Strongly
4) does not seem to be very reputable. (R)

 

Agree Agree Agree Undecided Disagree Disagree Disagree
Strongly Somewhat Somewhat Strongly

57

5) In my opinion is not a very selfish individual.

 

 

 

 

 

 

Agree Agree Agree Undecided Disagree Disagree Disagree
Strongly Somewhat Somewhat Strongly
6) Most people would be better off not being associated with . (R)
Agree Agree Agree Undecided Disagree Disagree Disagree
Strongly Somewhat Somewhat Strongly
7) gives the impression they are just.
Agree Agree Agree Undecided Disagree Disagree Disagree
Strongly Somewhat Somewhat Strongly
8) Not many individuals are as honest as
Agree Agree Agree Undecided Disagree Disagree Disagree
Strongly Somewhat Somewhat Strongly
9) is not an individual I would prefer to be like. (R)
Agree Agree Agree Undecided Disagree Disagree Disagree
Strongly Somewhat Somewhat Strongly
10) I would not trust to tell the truth when required to. (R)
Agree , Agree Agree Undecided Disagree Disagree Disagree
Strongly Somewhat Somewhat Strongly

 

Note. Reverse scoring should be performed for items with (R).

58

APPENDIX F

Information Evaluation Instrument
Instructions: On the items below please think about the final piece of information that

was revealed by in the scenario you just read. For each item, circle

 

the number between the adjectives which best represents your feelings about that final
piece of information relative to that pair of adjectives. Numbers ‘1’ and ‘7’ indicate a
very strong feeling. Numbers ‘2’ and ‘6’ indicate a fairly weak feeling. Number ‘4’
indicates you are undecided or do not understand the adjectives themselves. If you have
any questions please ask the researcher.
Valence
Good 1 : 2 : 3 : 4 : 5 : 6 : 7 Bad (R)
Detrimental 1 : 2 : 3 : 4 : 5 : 6 : 7 Beneficial
Positive 1 : 2 : 3 : 4 : 5 : 6 : 7 Negative (R)
Unfavorable l : 2 : 3 : 4 : 5 : 6 : 7 Favorable
Destructive 1 : 2 : 3 : 4 : 5 : 6 : 7 Constructive

Advantageous 1 : 2 : 3 : 4 : 5 : 6 : 7 Disadvantageous (R)

 

Note. Reverse scoring should be performed for items with (R).

59

APPENDIX G

Induction Check Instrument ’

Instructions: On the items below, without going back and referring to the scenario you

just read, please think about what you read and answer the questions by responding on the

blank space provided. If you have any questions please ask the researcher.

1)

2)

3)

4)

5)

6)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

What was the final piece of information revealed by ?

Could the final piece of information that revealed been revealed

by someone other than ?

If so, was aware the information could have been revealed by

someone else?

Did give an explanation for the final piece of information they

revealed?

If so, what was ’s explanation for the final piece of information

they revealed?

It was easy for me to imagine myself in the position of

Agree Agree Agree Undecided Disagree Disagree Disagree

Strongly Somewhat Somewhat Strongly

60

APPENDIX H

Demographics Instrument
Instructions: On the items below please answer the questions by responding on the blank
space provided. If you have any questions please ask the researcher.

1) What is your age?

 

2) What is your sex?

 

3) How many years have you been in college?

 

4) Where are you from? (City, State)

 

61

REFERENCES

62

REFERENCES

Asch, S. E. (1948). The doctrine of suggestion, prestige, and imitation in social
psychology. Psychological Review, 55, 250-276.

Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in
social psychological research: Conceptual, strategic, and statistical considerations.
Journal of Personality and Social Psychology, 51, 1 173-1 182.

Berlo, D. K., Lemert, J. B., & Mertz, R. J. (1970). Dimensions for evaluating the
acceptability of message sources. Public Opinion Quarterly, 33, 563-576.

C ialdini, R. B. (1993) Influence: Science and practice. New York, NY: Harper Collins.

Dolnik, L., Case, T. I., & Williams, K. D. (2003). Stealing thunder as a courtroom tactic
revisited: Processes and boundaries. Law and Human Behavior, 27, 265-285.

Eagly, A. H., Wood, W., & Chaiken, S. (1978). Causal inferences about communicators
and their effect on opinion change. Journal of Personality and Social Psychology,
36, 424-435.

Erber, R., & Erber, M. W. (2001). Intimate Relationships: Issues, theories, and research.
Needham Heights, MA: Allyn & Bacon.

Hovland, C. I., Janis, I. L., & Kelley, H. H. (1953). Communication and Persuasion:
Psychological studies of opinion change. New Haven, CT: Yale University Press.

J onides, J. (1995). Working memory and thinking. In E. E. Smith & D. N. Osherson
(Eds.) Thinking: An invitation to cognitive science, vol. 3, (215-265). Cambridge,
MA: The MIT Press.

Kelley, H. H. (1950). The warrn-cold variable in first impressions of persons. Journal of
Personality, 18, 431-439.

Kelman, H. C., & Hovland, C. I. (1953). “Reinstatement” of the communicator in
delayed measurement of opinion change. Journal of Abnormal and Social

Psychology, 48, 327-335.

Kunda, Z. (1999). Social Cognition: Making sense of people. Cambridge, MA: MIT
Press.

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

63

McCroskey, J. C., Hamilton, P. R., & Weiner, A. M. (1974). The effect of interaction
behavior on source credibility, homophily, and interpersonal attraction. Human
Communication Research, 1, 42-52.

McElhaney, J. W. (1987). Stealing their thunder. Litigation, 13, 59-60, 68.

McGuire, W. J. (1964). Inducing resistance to persuasion: Some contemporary
approaches. In L. Berkowitz (Ed.) Advances in Experimental Social Psychology,
vol. I, (191-229). San Diego, CA: Academic Press.

O’Keefe, D. J. (2002). Persuasion: Theory and research, Thousand Oaks: Sage.

Ondrus, S. A., & Williams, K. D. (May, 1995). Effects of Stealing Thunder by a Political
Candidate: Admit or deny? Paper presented at Midwestern Psychological
Association, Chicago, IL.

Osgood, C. E., Suci, G. J., & Tannenbaum, P. H. (1957). The Measurement of Meaning.
Urbana, IL: University of Illinois Press.

Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of persuasion. In
L. Berkowitz (Ed.), Advances in Experimental Social Psychology, vol. 19, (123-
205). San Diego, CA: Academic Press.

Schwarz, N. (1995). Social cognition: Information accessibility and use in social
judgment. In E. E. Smith & D. N. Osherson (Eds.) Thinking: An invitation to
cognitive science, vol. 3, (345-376). Cambridge, MA: The MIT Press.

Walster, E., Aronson, E., & Abrahams, D. (1966). On increasing persuasiveness of a low
prestige communicator. Journal of Experimental Social Psychology, 2, 325-342.

White, P., & Williams, K. D. (August, 1998). Does Stealing Thunder Always Work?:
Black candidates beware! Paper presented at the American Psychological
Association Annual Convention, San Francisco, CA.

Williams, K. D., Bourgeois, M. J ., & Croyle, R. T. ( 1993). The effects of stealing thunder
in criminal and civil trials. Law and Human Behavior, 1 7, 597-609.

Williams, K. D., & Dolnik, L. (2001). Revealing the worst first: Stealing thunder as a
social influence strategy. In J. P. Forgas and K. D. Williams (Eds), Social
Influence: Direct and indirect processes. Philadelphia, PA: Psychology Press.

Zablocki, K. (1996). Stealing Thunder About Having a Sexually- Transmitted Disease.
Unpublished Honor’s Thesis. University of Toledo, OH.

64

   

lllililjjlllijflljlllllIfljljllllll