ABSTRACT
THE EFFECTS OF TRUTH AND DESIRABILITY

EVIDENCE ON JUDGEMENTS OF THE TRUTH
AND DESIRABILITY OF A PROPOSITION

By
Ted J. Smith III

One of the most basic precepts of rhetorical theory is that evidence
should have a strong and reliable impact on belief. The vast majority
of the empirical studies designed to test this precept have, however,
produced only negative or conflicting results. It was this problem
which gave rise to the present study. Our inquiry began with a critical
review of both rhetorical theory and previous evidence research. On the
basis of this review we identified a number of theoretical, definitional
and methodological shortcomings of the earlier research which may have
contributed to the disappointing results of that research. we then devel-
oped a new model of the relation of evidence to belief, grounded primarily
in a set of distinctions drawn from.the rhetorical theories of Aristotle
and Richard Weaver, which seemed capable of overcoming those shortcomings.
In our research we tested four hypotheses derived from the model and two
hypotheses suggested by earlier empirical research. Specifically, we
sought to determine: (1) the effects of truth and desirability evidence
on rational and desirable belief; (2) the effects of training in evi-
dence usage and argumentation on the link between rational and desirable
belief; and (3) the effects of general education level on the link

between rational and desirable belief.

Ted J. Smith III

Our research began with the construction of six evidencebased mes-
sages. Three of the messages supported a single experimental proposition
and three opposed the same proposition. One of each type of message was
supported by no evidence statements, one of each was supported by desira-
bility evidence statements, and one of each was supported by truth evi-
dence statements. Nine experimental treatments were then constructed
using all possible combinations of the two types of messages. In Phase 1
of the study these treatments were randomly distributed to 273 under-
graduate students who read the messages and completed scales measuring
their estimates of the truth and desirability of the proposition. Several
weeks later, after the students had read, discussed and applied material
on the proper structure of arguments and the use of evidence in argumen-
tation, the same nine treatments were again randomly distributed to 166
undergraduates, most of whom had participated in the earlier manipulation.

The first two experimental hypotheses, which predicted rank orders
of rational belief and desirable belief ratings, were largely supported
by the data. The next two hypotheses, which predicted the effects of
training and education level on the correlation between rational and
desirable belief, were not supported. The last two hypotheses, which
predicted rank orders of rational and desirable belief ratings for the
Phase 2 subjects, also were not supported.

The findings were interpretted as generally supportive of the pro-
posed model, although three potential modifications to the model were
discussed. A number of research extensions suggested by the results of

this study were also discussed.

THE EFFECTS OF TRUTH AND DESIRABILITY
EVIDENCE ON JUDGEMENTS OF THE TRUTH
AND DESIRABILITY OF A PROPOSITION

\ BY

  
 

Ted TLH‘Smith, III

A THESIS

 
 

Submitted to
Michigan State U versity
in partial fulfillment o the requirements
for the degr e of

Department of Communi ation

1972

x‘t
9
IX

Accepted by the faculty of the Department of
Comunication, College of Commication Arts, Michigan

State University, in partial fulfillment of the

 
    
 

   

Guidance Committee:

 
  
  

‘1" \a‘éds Chajmn
[MM/.1

1!. 11/.
”5 “040,.

 

 

ACKNOWLEDGEMENTS

It is of course impossible to thank all those persons who helped
me to reach this point in my academic career. It seems appropriate,
however, to begin such an attempt by acknowledging the tireless assis-
tance of Prof. Donald P. Cushman, who directed this thesis. Whatever
capabilities I may possess as a scholar of human communication are due
in large part to his efforts. More than giving me just guidance and
stimulation, he introduced me to the true art of thinking.

My guidance committee members, Dr. Gerald R. Miller and Dr. Everett
M. Rogers, were constant sources of aid and encouragement, both during
the execution of this thesis and throughout:my stay at Michigan State
University.

My special thanks go to Dr. Joseph Woelfel, who provided vitally
needed assistance during the final stages of the thesis, and Dr. Bradley
S. Greenberg, who was instrumental in providing me the opportunity to
continue my studies in the field of communication.

Last, but most important, I thank my wife, RoseMary, to whom I
dedicate this thesis. It was her sacrifices, encouragement and help

that made the thesis possible, and her love that makes it meaningful.

ii

..—_ —’"_"—Y

 

TABLE OF CONTENTS

ACKNOWLEDGEMENTS . . . . . . . . . . . . . . . . . .

LIST OF 'I|ABI‘m O O O O O O O O O O O O O O O O O O 0

LIST OF FIGURES . . . . . . . . . . . . . . . . . .

Chapter

I.

II.

INTRODUCTION 0 O O O O O O O O O O O O O O

Purposes of the Study . . . . . . . .
Divisions of the Thesis . . . . . . .
Statement of the Problem . . . . . . .
Inadequate Theoretical Formulations .
Inadequate Conceptual and

Operational Definitions . . . . . . .
Inadequate contrO]. o o o o o o o o o o 0
Theoretical Reformulation . . . . . . .
Reformulation of Conceptual Definitions
Reformulation of Control Procedures . .
Overview of Research Design

and Procedures . . . . . . . . . . .
Rationale for Hypotheses . . . . . . . .

WTHODOIDGY O O O O O O O O O O O O O O O O

PreteStOOOOOOOOOOOOOOOO
samples OOOOOOOOOOOOOOOO
Operationalization of Independent
VariableS..............
Operationalization of Dependent
variablesoooooooooo
Questionnaire Construction . .
Data Collection Procedures . .
Control Procedures . . . . . .
DeSignoooooooooooo

iii

ii

vi

Page

—l

Table of Contents (Cont'd.)

Chapter Page
III 0 RESULTS 0 O O O C O O C O O O O O O O O O O O O O O O O 45
Effects of Evidence: Phase I . . . . . . . . . . . . 45

Effects of Education Level and Training . . . . . . 48

Effects of Evidence: Phase II . . . . . . . . . . . 49

IV 0 DISCUSS ION O O O O O O O O O O O O O O C O O O O O O C 53

SW 0 O O O O C O O O O O O O O O O O O O O O O 53
Effects of Evidence on Belief:

Conclusions . . . . . . . . . . . . . . . . . . . 55
Effects of Outside Variables:

Conclusions . . . . . . . . . . . . . . . . . . . 60
Research Implications . . . . . . . . . . . . . . . 62

REmR-ENCES O O O O O O O O O O O O O O O O C O O O O O O I O O O 64
APPMD H A O O O O O O O O O O O O O O O O O O O C O O O O O O O 68

APPENDIX B O O O O O O O O O O O O O O O O O O O O O O O O O O O 75

iv

Table

\IOU‘cl-‘w

LIST OF TABLES

Page

Comparison of "evidence" and "no evidence"
treatments 0 O O O O O O C O O O O O O O O O O O O O O 1 4
Evidence sources and their qualifications . . . . . . . . 42

Rational belief ratings for Phase I subjects . . . . . . 46
Desirable belief ratings for Phase I subjects . . . . . . 47
Belief correlations . . . . . . . . . . . . . . . . . . . 49
Rational belief ratings for Phase II subjects . . . . . . 50

Desirable belief ratings for Phase II subjects . . . . . 51

Figure

Ull-‘UJ

LIST OF FIGURES

Example of semantic differential battery . .
Nbdel of the relation of evidence to belief
Experimental design . . . . . . . . . . . .
Dependent measurement instruments . . . . .

Full experimental design with sample sizes .

Page
16
19

37

CHAPTER I

INTRODUCTION

Historically, rhetoric has been understood to operate as the ground
for communication in the quest of truth (Grimaldi, 1958). Rhetoric,
then, by its very nature gives careful attention to the rational aspects
of proof. To build a rational proof, speakers need evidence to establish
premises and lay down foundations for statements from.which inferences
can be made. In this context, theories of evidence are largely prescrip-
tive, in that statements about evidence are cast in terms of the valid-
ity of arguments. Theories of rhetoric traditionally have held that the
proper use of evidence is central to establishing belief in the validity

or probable truth of a proposition.

Purposes of the Study
This study was designed to fulfill three major purposes: (1) to

determine the effects of different forms of evidence on judgements of
the truth and desirability of a proposition; (2) to determine the
effect of training in evidence usage and argumentation on the correla-
tion between judgements of the truth and desirability of a proposition;
and (3) to determine the effects of general education level on the cor-
relation between judgements of the truth and desirability of a proposi-

tion.

Divisions of the Thesis

The thesis will be divided into the following four chapters:
(1) The remainder of this chapter will be devoted to (a) a statement
of the problem which gave rise to this study, (b) a critical evalua-
tion of previous empirical research dealing with the relationship
between evidence and belief, (c) a theoretical and operational reforms
ulation of this relationship, and (d) a statement of the hypotheses
tested in this study. (2) Chapter II will discuss the methodology
employed in this study. (3) Chapter III will present the results of‘
our inquiry. (A) Chapter IV will discuss conclusions that may be

drawn from the study and make recommendations for future research.

Statement of the Problem

While rhetorical theory provides a significant rationale for the
relationship between evidence and belief, the results of a number of
empirical studies have cast doubt on the validity of this traditional
position. Taken as a whole, the empirical studies investigating the
relation of evidence to belief have either (1) reported that the impact
of evidence on belief is insignificant, or (2) been conflicting in their
results. The following is a brief summary of the conclusions of this
research.

First, several studies investigated the effects of evidence-plus-
assertion and assertion-only speeches on attitude change. Cathcart
(1955), Bostrom and Tucker (1969) and Kline (1969) found that a speech
containing assertions supported by evidence is more effective in

changing attitudes than a speech containing only generalizations and

3
assertions. On the other hand, Costley (1958) and Wagner (1958) found
no significant differences in attitude change between the same two
types of speeches.

Second, several studies investigated the effects of evidence-plus-
assertion-plus-qualified-authority and assertion-only speeches on atti-
tude change. Cathcart (1955) and Bostrom.and Tucker (1969) found that
speeches containing evidence attributed to qualified authorities were
significantly more effective in changing attitudes than speeches con-
taining only unattributed assertions. Ostermeier (1967) and Whitehead
(1971), however, found no significant differences in attitude change
between the same two types of speeches. McCroskey (1967, 1970) reported
that evidence-plus-assertion-plus-qualified-authority speeches were
significantly more effective in changing attitudes than assertion-only
speeches when the speaker was originally perceived as low or moderately
credible, but were not significantly more effective when the speaker
was perceived as highly credible. One researcher, Bettinghaus (1953),
discovered that speeches containing evidence-plus-assertion-plus—
qualified-authority produced significantly'more attitude change than
speeches containing only evidence-plus-assertion, but his findings have
never been replicated. Cathcart (1955), Gilkinson, Paulson and Sikkink
(1954), Sikkink (1956) and Bostrom and Tucker (1969) all found no signi-
ficant differences between the same two types of speeches. 'Further,
Bostrom and Tucker (1969) reported that a speech containing evidence-
plus-assertion-plus-authority was significantly less effective in chang-
ing attitudes than speeches containing evidence-plus-assertion-plus-

authority-and-qualifications (i.e., the "qualified authority" treatments

4

discussed above) or evidence-plus-assertion alone. This confirmed a
similar, but non-significant, trend in the 1955 Cathcart study.

Third, several studies investigated the effects of speeches con-
taining high and low quality evidence on attitude change. warren
(1971) found that speeches containing testimony attributed to highly
credible authorities produced significantly more attitude change than
speeches in which the same testimony was attributed to authorities of
low credibility. Dresser (1963) and Gardner (1966) found no signifi-
cant differences in attitude change between messages containing evi-
dence rated as high in quality by a panel of judges and those contain-
ing low quality evidence. These latter results are consistent with
the findings of a study be Harte (1971) in which he discovered that:
"...audiences are not notably successful at applying the appropriate
tests of evidence to material offered as proof of an assertion" (p.112).
A Wagner (1958) and Ostermeier (1967) manipulated the amount of evidence
in experimental messages but found no significant differences in the
resulting attitude change scores.

Fourth, several studies investigated the effects of evidence on
variables other than immediate attitude change. MbCroskey (1967)
found that in a number of his experiments a source who was initially
perceived as low-to-moderate in credibility was rated significantly
higher in credibility if he included evidence in his speech than if he
did not. This finding has been at least partially supported by the
studies of Ostermeier (1967) and Whitehead (1971). Next, McCroskey
(1967) also discovered that the inclusion of evidence in a persuasive

message resulted in significantly greater delayed attitude change,

5
regardless of the initial credibility of the source. A final discov-

ery by McCroskey (1967) was that evidence served as an effective
inhibitor to immediate counterpersuasion attempts.

In the great majority of the above studies, the experimenters
asked whether, as proof becomes "better" in the prescriptive sense,
it also becomes more effective in changing the attitudes or beliefs of
an audience. The findings of these studies are sufficiently conflicting
or negative to demand a careful re-examination of the relationship
between evidence and belief. A detailed review of rhetorical theory
and the prior empirical research suggests three general factors which
may account for these findings. They are: (1) an inadequate theoreti-
cal formulation of the relation of evidence to belief; (2) inadequate
conceptual and operational definitions of evidence and belief; and (3)
inadequate control over certain critical intervening variables. We

will now discuss each of these factors in detail.

Inadequate Theoretical Formulations

A careful review of rhetorical theory suggests at least three
sources of problems in the previous empirical research investigating
the relation of evidence to belief. These are: (1) a failure to dis-
tinguish between different types of belief; (2) a failure to distin-
quish between different types of evidence; and (3) a failure to deal
with the theoretically posited and empirically verified link between
rational belief and desirable belief.

Since Aristotle's time, rhetoric has been viewed as the counterpart

of dialectic. As such, it aims at securing judgements in the realm of

6
the contingent. Thus at the very center of rhetorical theory is the
idea of the audience as judge, as that which is to exercise the act of
judgement. Rhetoric does not effect persuasion in the sense of manip-
ulation, but rather it creates an atmosphere wherein the judge or audi-
ence may make its own decision. This prescription has been well articu-
lated by Grimaldi (1958, p.374):

The art or technique of the rhetorician is to perceive and

present those things which make decision, and a definite

decision, possible, but to stop with the presentation. The
audience at this point must come in to accept or reject,

to make its particular judgement to execute or refrain

from action. Rhetoric, then, is preparatory for action.

In this context, rhetorical theory distinguishes between proof
which argues for the truth of a given proposition by presenting evi-
dence in favor of the probable truth of the premises from which the
proposition flows, and proof which seeks to establish the desirability
of a given proposition by appealing to the passions or desires from
which support of the proposition flows (Aristotle, 1941, pp.1329 -
1330). Although the distinction is commonplace, its full implications
have generally been ignored. It is critical to note that so-called
rational and emotional proofs differ not only in form.but also in
intent. Rational proof seeks to establish rational belief in the
proposition. That is, it aims at establishing the validity or truth
value of a proposition. Emotional proof seeks to establish what we
will call desirable belief, or belief in the desirability of the
proposition or its consequences. Researchers investigating the rela-

tion of evidence to belief have failed theoretically and operation-

ally to make the distinction between rational and desirable belief.

7
That is, while they have tried to manipulate only the elements of logi-
cal proof, they have invariably employed some generalized measure of
belief (e.g., Likert scales, semantic differentials) in assessing the
impact of those manipulations. As a result, judgements regarding the
truth value of a proposition are confused with general attitudes toward
the acceptance of the proposition, and the effects of rational proof
are suppressed. This is especially disturbing because empirical sup-
port for the validity of this distinction may be found in research
reported by McGuire (1960b). McGuire, while investigating the persua-
sive effects of dissonance-producing messages, was able to account for
more than 90% of the variance in attitude change by measuring just two
variables: (1) the probability that a set of propositions were true,
and (2) the desirability of the consequences of those propositions.

The second source of problems is that previous researchers have
failed to distinguish between different types of evidence. In addition
to the basic distinction between rational and emotional proofs, rhetor-
ical theory also distinguishes between different types of emotional
proofs. Aristotle suggests (191.1, pp.597 - 599) that emotional proof
can effect persuasion through either (1) appeals to appetitive desire,
or (2) appeals to intelligence or reason in the service of desire.

The first type of appeal is the one Aristotle treats so contemptuously
in the Rhetoric (1941, p.1330). As for the second, Cushman (1968)
concludes after an examination of Aristotle's broader philosophical
position that "The standard of excellence in the functioning of emo-
tional proof is the appeal to intelligence in the service of desire"

(p.5). For our purposes, the critical point of the distinction is that

8
the second type of emotional proof requires the use of evidence to
establish the rational desirability of a proposition or its conse-
quences. Thus we are forced to differentiate between evidence which
argues for the truth of a proposition by establishing the truth or
probable truth of the premises from which it flows (truth evidence)
and evidence which argues for the desirability of a proposition or its
consequences (desirability evidence). Previous research on emotional
proofs has failed to make the distinction between the two types of
emotional appeals. Both the research on logical versus emotional
appeals (summarized by Becker, 1963) and so-called fear appeals (sump
marized by Miller, 1963, and updated by'Pfiller and Hewgill, 1966)
have invariably mixed together elements of the two types of emotional
appeals in their manipulations. Research on the relation of evidence
to belief, which developed independently, has failed to make the dis-
tinction both between different types of emotional appeals and, result-
ingly, between truth evidence and desirability evidence. Instead,
evidence has been treated as a purely unitary concept, appropriate
only in the context of rational proof. Classifications of evidence have
been made solely on the basis of structural criteria (e.g., fact versus
opinion, presence or absence of the qualifications of the source, logi-
cal consistency) rather than on the basis of the type of appeal sup-
ported. This failure is especially disturbing because we would expect
desirability evidence to have a different impact on general attitude
than truth evidence, both because desirability evidence would seem to be
more powerful than truth evidence under certain conditions (as will be

argued more fully below) and because the contributions of rational and

9

desirable belief to general attitude are unclear. In short, the fail-
ure of previous researchers to distinguish between the different types
of evidence, and their inclusion of the two types of evidence in single
appeals, makes it impossible to either accurately predict or interpret
the impact of their experimental messages on attitudes.

The third source of problems is the failure of previous evidence
researchers to take account of the theoretically posited and empirically
verified influence of desirable belief on rational belief. This idea (A
is implicit in both the rhetorical and philosophical works oanristotle !

l

(1941, pp.1403 - 1407 and 1025 - 1027) but has been made explicit and ‘
fundamental in the works of the modern rhetorical theorist Richard 1
Weaver. According to Weaver, man participates in culture at three
levels of conScious reflection: "...his specific ideas about things,
his general beliefs or convictions, and his metaphysical dream of the
world" (1948, p.18). Of these, the highest level is the metaphysical
dream, which Weaver sees as "...an intuitive feeling about the imminent
nature of reality, and this is the sanction to which both ideas and
beliefs are ultimately referred for verification" (1948, p.18). Further,
as Cushman and Hauser (In Press) have argued:

Each of these levels holds man as imposing something upon the

raw data of the world to order, shape and evaluate his exper-

ience of it. Confronted by raw data his initial reaction is

one of wonder. In this state where phenomena are without

order, reason is impossible. Rational activities commence

only after an appropriate perspective is gained: one which

orders and shapes the phenomena into a meaningful whole. This

perspective comes from one's sentiment toward the world. Sen-

timent draws us to the world. It is an affective interest.

Moreover, it is what man superimposes upon raw experience to

make it personally meaningful and fit material for rational
activity. For Weaver, sentiment is anterior to reason. (p.4)

10
Thus it is man's desires, ultimately embodied in his metaphysical
dreams, that give meaning to reality. This conceptualization has two
major implications for evidence research. First, it suggests that
factual evidence, in and of itself, should have little impact unless
it can be easily interpretted within the framework of the auditor's
metaphysical dream. Second, it suggests that desirability arguments,
if (and only if) they link with the auditor's metaphysical dream,
should influence not only his belief in the desirability of a propo-
sition but also his belief in the truth of the proposition, for it
is only through perceptions of desirability that truth acquires mean-
ing for him. In short, Weaver's position holds that belief in the
desirability of a proposition influences belief in the truth of that
proposition. Empirical support for the validity of this distinction
may be found in the research of McGuire (1960a,b) and Dillehay, Insko,
and Smith (1966). These researchers found, with one exception, that
'judgements of the truth of a proposition were determined in part by
judgements of the desirability of the proposition. The single excep-
tion was among upper level college students in one experiment in the
Dillehay, Insko and Smith study. These students seemed to be making
the judgements independently.

Although this distinction is critical to the theoretical formula-
tion underlying the present study, it would not be strictly necessary
in a study of the effects of truth evidence on rational belief. Pre-
vious evidence researchers, however, have not restricted themselves
solely to truth evidence and rational belief. Instead, they have

looked at the effects of evidence in general on attitudes in general.

11
Given this approach, the causal relationship between desirable belief
and rational belief does become critical. Let us assume that a given
subject's general attitude on a proposition is some composite function
of his rational belief and desirable belief on the proposition. If this
subject were then presented with truth evidence in favor of the proposi—
tion, we would expect his general attitude to change in direct relation
to the amount of change in rational belief induced by the evidence. If,
however, this subject were presented with desirability evidence on the
proposition, we would expect his general attitude to change both in
response to the change in desirable belief induced by the evidence and
in response to the change in rational belief induced by the change in
desirable belief. Thus, once again, to the extent that previous evi-
dence researchers failed to distinguish between different types of

evidence and belief it becomes impossible to interpret their results.

Inadequate Conceptual and Operational Definitions

Closely related to the problems which follow from the failure of
previous evidence researchers to draw the theoretical distinctions
noted above are those which follow from.the failure of these research-
ers to develop adequate conceptual and operational definitions of cer-
tain key concepts. In general, previous evidence research is marked
by failures to adequately conceptualize and operationalize (1) "evi-
dence" treatments, (2) "no evidence" treatments, and (3) belief.

As we argued above, one major theoretical shortcoming of prior
evidence research was the failure to distinguish between truth evidence

and desirability evidence. This, of course, represents a serious

12

conceptual problem as well. Even if we were to accept the earlier con-
ceptualizations of "evidence" treatments, however, we would still be
faced with several grave operational problems. In most previous stud-
ies, "evidence" treatments consisted of a verbal matrix in which a nume
ber of pieces of evidence were embedded. Each piece of evidence usually
consisted of three distinct parts: (1) the evidential data, (2) the
source of the evidential data, and (3) the qualifications of the source
of the evidential data. Thus each piece of evidence had at least separ-
ate and possible interacting sources of persuasive potential. In most
prior research it is impossible to be sure which of these various ele-
ments in the "evidence" treatments is or is not producing the persuasive
effect. Empirical support for the validity of this distinction may be
found in the research of Cathcart (1955) and Bostrom.and Tucker (1969).
These researchers discovered significant (but not necessarily cumula-
tive) differences in the amount of attitude change produced by speeches
containing only evidential data, data-plus-source, and data-plus-source-
plus-qualifications of the source.

A second, and hopefully more restricted, problem with previous
operationalizations of "evidence" treatments is the use of evidence
that does not logically support the proposition. For example, Kline
(1969), in studying the effects of "specific", "nonspecific" and "no"
evidence treatments on attitude change, used the following piece of
evidence as major support for the proposition "The popularity of clas-
sical music has increased over the past ten years". "Specific evidence:
Among all types of musical performers, two classical performing artists,

Leonard Bernstein and Arturo Rubenstein, are among the ten musicians

C

 

13
most in demand" (p.409). In this and all similar cases it would be
extremely difficult to interpret observed changes in the dependent
variable, especially if these changes were found to be non-significant.

"No evidence" treatments suffer from.similar definitional diffi—
culties. Actually the term "no evidence" is a misnomer because these
treatments retain, with only slight modifications, the essential compo-
nent of evidence--the data in the case of truth evidence, or the asser-
tion of desirability or undesirability in the case of desirability evi-
dence. The modifications usually consist of: (1) deletion of the
source's name and qualifications, (2) generalization of the data, or (3)
both of the above. To give a flavor for these manipulations, consider
Table 1. It compares three fairly representative samples taken from
actual "evidence" and "no evidence" treatments (NbCroskey, 1966, pp.

170 - 183).

The samples of "no evidence" treatments in Table 1 probably fail to
meet the logical or prescriptive tests of adequacy imposed by some audi-
ences. Nevertheless, these treatments do consist of data, and this
data does have the potential to induce attitude change. Thus it is
hardly surprising that the "no evidence" speeches from which these same
ples were taken produced significant changes in attitude in comparison
with no-message control groups (McCroskey, 1966, p.55). In other stud-
ies, messages containing "evidence" were compared solely with messages
containing "no evidence" with regard to attitude change. In such cases
a finding of no significance could simply reflect the failure to develop

an adequate operational definition of "no evidence" treatments.

14

 

Table 1.

Comparison of "evidence" and "no evidence" treatments

 

"Evidence" treatments

According to figures released
by the Department of Health,
Education and Welfare of the
Federal Government, over one-
third of the nation's high
schools do not offer such
essential college preparatory
subjects as chemistry or phys-
ics, and about the same number
don't even offer one foreign

language.

"No evidence" treatments

Many of our schools do not offer
such essential college preparatory
subjects as chemistry or physics,
and others don't even offer one

foreign language.

 

In their report entitled The
Financial Status of Public
Schools the committee on edu-
cational finance of the
National Education Associa-
tion reports that last year
the nation was short 118,000
qualified teachers just to
meet minimum standards.

Across the nation we are short
more than 100,000 teachers just
to meet minimum standards.

 

According to figures released
by the United States Office of
Education in January of this
year, 25.4% of the nation's
classrooms are, in their words,
"obsolete and unacceptable"
for public schools because of
such things as fire hazards.
...we find that over 375,000
classrooms are presently
unacceptable.

A very large percentage of the
classrooms presently in use are
obsolete and unacceptable for
public schools because of such
things as extreme fire hazards.
Over three hundred thousand of
our present classrooms are
unacceptable.

 

15

Turning next to the definitions of belief employed in previous
evidence research, we are again met with several problems. First, as
we noted above, the distinction between rational and desirable belief
has been systematically ignored, resulting in the use of such general-
ized measures of attitude change as the WOodward Shift-of-Opinion Ballot,
Likert scales and semantic differential batteries. This, we have argued,
has tended to suppress the effects of rational proof and is quite proba-
bly a contributing factor in the negative or conflicting findings of
earlier evidence research. A second and more specialized problem
arises, however, when the semantic differential is used as the dependent
measure of attitude change. Fortunately, only one researcher, McCroskey
(1967, 1970), has used the semantic differential extensively to measure
attitudes, although he has been by far the most prolific of all the
evidence researchers. The semantic differential was developed (Osgood,
Suci and Tannenbaum, 1957) in connection with studies of connotative
meaning. Specifically, it was devised to get some measure of the hypo-
thesized rm...sm bonds in Osgood's mediational learning hypothesis. A
semantic differential scale is usually composed of seven points bounded
by a pair of polar adjectives. These adjective pairs, when used to
rate an object, concept or proposition, are found to group into three
primary factors--evaluative, potency and activity. Although in atti-
tude measurement primary concern is with the evaluative dimension, use
of all three factors generally results in higher predictability of
attitude change. When semantic differential scales are used to measure
the effects of evidence, two questions become critical. First, one

must ask whether the semantic differential is capable of reliably

16

 

Military activities in South Viet Nam should be
significantly expanded.

 

 

 

 

Good : : : : : : Bad
Weak : : : : : : Strong
Fast : : : : : Slow
Worthless : : : : : : valuable
Pleasant : : : : : : Unpleasant

 

 

Figure 1. Example of semantic differential battery

 

detecting changes in rational belief. We have addressed this question
previously. Second, one must ask whether the specific collection of
dimensions operationalized are capable of tapping relevant antecedents
in the message. To illustrate the issue we are raising, we have cone
structed a semantic differential battery in Figure 1 similar to those
employed by McCroskey (1967, 1970) in his evidence research. Our argu-
ment is that certain pieces of evidence employed in support of the
experimental proposition may not seem to induce attitude change because
of the failure of the evidence to link with the proposition in such a
manner as to involve the dimensions of evaluation cited. For example,
desirability evidence supporting expanded military activity as the only
"honorable" course of action might produce significant attitude changes
that were not detected by the measurement instrument shown in Figure 1.
In such cases, a finding of "no significance" would simply reflect a
defeat in either the operational definition of evidence, the operational

definition of belief, or both.

17
Inadequate Control

In addition to the theoretical and definitional problems that we
have outlined, the great majority of previous evidence studies failed
to exercise adequate control over the following variables: (1) source
credibility, (2) familiarity with the evidence, and (3) redundancy.

McCroskey (1969) has argued that source credibility interacts with
evidence to account for the conflicting results apparent in previous
evidence research. MCCroskey's research demonstrates that a high cred-
ibility source produces as much or more attitude change from.an asser—
tion unsupported by evidence as a low credibility source who supports
his assertion with evidence. MbCroskey concludes:

Including good evidence has little, if any, impact on
immediate audience attitude change or source credibility

if the source is initially perceived to be high-credible.

Including good evidence may significantly increase

immediate audience attitude change and source credibility

when the source is initially perceived to be moderate-to-

low credible, when the message is well delivered, and

when the audience has little or no prior familiarity with

the evidence. (1969, p.175)

Since source credibility has seldom been explicitly controlled, espe-
cially in the earlier evidence studies, we must conclude with MbCroskey
that differences in levels of source credibility may have interacted
with evidence to produce the conflicting results of previous evidence
research.

Arguing from a dissonance theory position, McCroskey (1967, 1969)
has also proposed that if subjects were aware of evidence before it was
presented by a speaker, the evidence might not have a significant
impact on either their attitudes on the topic or their perceptions of

the source. The results of a single study conducted by MbCroskey (1967)

 

 

18
suggest that this may indeed be the case. Thus another possible
source of confusion in the evidence treatments of previous studies
is the failure to control for audience familiarity with the evidene
tial data so that its persuasive effects can be measured free from
other influences.

Finally, in several of the evidence studies reviewed above, the
researchers attempted to control for message length by repeating or
restating assertions in the "no evidence" treatments. In most of the
remaining studies, the messages containing evidence were somewhat
longer than the "no evidence" messages (cf., Table 1, above). Although
the effects of minor variations in message length are unknown, attene
tion has been directed recently to the effects of redundancy on atti-
tude change. For example, Troldahl, Costello and Robeck (1969) have
reported that redundant messages produce more attitude change than
non-redundant messages. This suggests that in those studies where
redundancy was not explicitly controlled, redundancy effects may have
been confounded with the effects of evidence.

In summary, our analysis of previous research suggests the need
for reformulations of (1) the theoretical bases for examining the rela-
tion of evidence to belief, (2) the conceptual and operational defini-
tions of evidence, no evidence and belief, and (3) control procedures.
The next three sections of this paper will be devoted to these reforms

ulations.

19

Theoretical Reformulation

Our theoretical reformulation is concerned with the relation of
evidence to belief. It derives primarily from.the theoretical dis—
tinctions that were discussed in some detail above. These are: (1)
the distinctions between rational belief and desirable belief; (2) the
distinction between truth evidence and desirability evidence; and (3)
recognition of the impact of desirable belief on rational belief. Come
bining these distinctions we have developed the model of the relation

of evidence to belief which is reproduced in Figure 2

 

TE = Truth Evidence

DE = Desirability Evidence
TE ”"35 Er" RB = Rational Belief

DB = Desirable Belief

a 1 ' Erb = Error: Rational Belief
DE ATDBVC Edb Edb = Error: Desirable Belief

 

 

Figure 2. Model of the relation of evidence to belief

 

Several features of the model bear comment. First, we have incor-
porated the traditional rhetorical argument that the use of (truth)
evidence is central to establishing belief in the validity or truth
value of a proposition (i.e., rational belief). We have, however,
rejected on both theoretical and empirical grounds the notion that the
aaly way in which rational belief can or should be affected is through
the use of truth evidence. Second, by including the error terms (Erb
and (Edb) we explicitly acknowledge the effects of variables other than

evidence on belief, some of which may be more powerful than evidence.

20
In short, our purpose is not to create a general model of persuasion,
but rather to create a model of the persuasive impact of evidence.
Third, by including the relationship between desirability evidence and
desirable belief, we hope to focus attention on an area of evidence
usage that has not yet been systematically explored (although some
recent work has been done in the area by Clark and Hynes, 1970, and
Infante, 1971).

Before turning to our conceptual definitions, we must first meet
an obvious set of objections to our position. Ever since the attacks
of Woolbert (1917) and Yost (1917) on Winans' (1915) theory of persua—
sion, it has been fashionable to reject any rhetorical distinction
between "logic" and "emotion" as a return to faculty psychology. More
recently, adherents of this position have pointed to Becker's (1963)
conclusion that research on logical and emotional proofs has been
generally unproductive and Ruechelle's (1958) discovery that subjects
in his experiment were unable to distinguish between logical and emo-
tional appeals as proof of the invalidity of this type of distinction.
Without wishing to become embroiled in this controversy, we would
emphasize the following three points. First, while rational belief
and desirable belief are seen as largely independent, we have acknow-
ledged their interdependence by including the non-reciprocal link
between them in our model. We therefore claim a theoretical position
somewhere between the extremes in the controversy. Second, there is
empirical, as well as theoretical, support for the validity of our
distinction between rational belief and desirable belief. Studies by

McGuire (1960a,b), Dillehay, Insko and Smith (1966), and Cushman and

21

Smith (In Preparation) all indicate that the distinction is both valid
and scientifically interesting. Third, there is ample empirical and
theoretical support for the validity of our distinction between truth
and desirability evidence. Cronkhite (1961), using "logical" and "emo-
tional" appeals that are remarkably similar to our truth and desirabi-
lity treatments, found that subjects were consistently able to differ-
entiate between them. In a study related to the present one, Filion
(1972) found that subjects were able to reliably categorize statements
as truth evidence, desirability evidence or no evidence. Finally, as
Cronkhite (1964) has pointed out, Ruechelle's oft-quoted finding con-
sisted of nothing more than a failure to disprove the null hypothesis,

and thus was doomed from the start.

Reformulation of Conceptual Definitions
In this section, we will develop conceptual definitions for the

_four concepts in our model. Operational definitions for these terms
will be reported in Chapter II. In general, we will define evidence

as anythiagloffered 91 a spaaker or writer which is intendaggto support
agproposition. Using this as a base, we would then define truth evi-
dence as anything offered by a speaker or writer which is intended to
support the truth of a proposition, and desirability evidence as any-
thing offered by a speaker or writer which is intended to support the
desirability of a proposition. Two observations about these definitions,
which were strongly influenced by Miller's (1966) definition of evi-
dence, are in order. First, the term "anything" has been purposefully

chosen to avoid the plethora of empirically indefensible prescriptions

22
as to what constitutes "good" or "bad", "adequate" or "inadequate" evi-
dence. In our formulation, personal assertions of fact or value, quo-
tations from.authorities, logical deduction and skillful analysis, and
so on, are all considered as possible forms of evidence. Our sole
criterion for designating something as evidence is whether the speaker
or writer intends it to support the proposition in question. Second, the
word "support" is used in our definition in the sense of "providing a
foundation for" the proposition. That is, the speaker or writer must
intend for his manipulation to provide a foundation or basis for the
proposition to which the message recipients may refer in making their
judgements on the proposition.

Although the literature abounds with definitions of evidence, the
exact reverse is true of definitions of belief. Of the works reviewed
for this study, the vast majority simply left belief conceptually unde-
fined. 0f the remainder, some (e.g., Bettinghaus, 1966) merely elabo-
rate on dictionary definitions. Others (e.g., Kruger, 1967) provide
prescriptive definitions of what certain forms of belief should be, or
how they should be attained. A very few (e.g., Miller, 1966) report
or develop definitions which, while philosophically sound, seem too
broad to link well with our operationalizations. Given the abstract-
ness and complexity of the concept involved, this collective approach
is understandable and, perhaps, unavoidable. Any definition of belief
can be expected to draw criticism from at least some quarter. Regard-
less, we have developed a set of definitions which, for our purposes,

seem.apposite. In general, we will define belief as a;feeliag‘of vaga-

able certainty with ragard to a pgopgsition, manifested in one or more

 

5-. _

I .

23
jaagaaents on the pgopgsition. Using this as a base, we will define
rational belief as a feeling of variable certainty with regard to the
truth of a proposition, manifested in a judgement on the truth of the
proposition. Desirable belief is defined as a feeling of variable
certainty with regard to the desirability of a proposition, manifested

in a judgement on the desirability of the proposition.

Reformulation of Control Procedures
Because several variables tend to interact with evidence and

belief and thus bias experimental results, the following controls were
introduced into the study. More precise details of the control proce-
dures followed will be reported in Chapter II. First, in order to
control for the effects of varying levels of source credibility on
belief, written messages with no specific source attribution were used.
General attribution was to the Department of Communication at Michigan
State University, which was assumed to be moderate-to-high in credibi-
lity. Second, to control for the effects of audience familiarity with
the evidence, fictional data were used in all messages. Third, to
control for the effects of message length and redundancy, all messages

were of approximately the same length and noneredundant.

Overview of Research Desiga and Procedures
In the final section of this chapter we will report the hypotheses

 

tested in this study and their derivations. First, however, it would
be helpful to briefly describe our research design and some of the more

salient experimental procedures followed. The study was in the form of

 

24

 

 

 

 

 

 

 

 

Attack
No Des. Truth
No 1 2 3
Constructive Des. 4 5 6
Truth 7 8 9

 

 

Figure 3. Experimental design

 

a 3x3, replicated, after-only design. The primary independent variable
was evidence; the primary dependent variables were rational belief and
desirable belief. The study was executed in two phases. All subjects
in both phases of the study were undergraduate students enrolled in the'
basic Communication course at Michigan State University. In the first
phase, each subject was given an experimental message consisting of a
proposition and two one-paragraph arguments--one supporting the propo-
sition (Constructive Argument) and one opposing the proposition (Attack
Argument). Three forms of both the Attack and Constructive arguments
were identified, based on the type of evidence (i.e., truth evidence,
desirability evidence or no evidence) used to support the argument.

All possible combinations of Attack and Constructive arguments were
employed, thus generating the nine-cell treatment matrix shown in
Figure 3. After reading the experimental message, each subject then
rated the truth and desirability of the proposition, justified his
ratings in a one-paragraph essay, and answered a series of questions

on demographic variables.

25

Several weeks after completion of Phase I data collection, all
students in the basic Communication course were given training in the
structure of arguments (following the model proposed by Toulmin, 1959)
and the proper use of evidence, as part of their required coursework.
This seemed to present a unique opportunity to study the effects of
this type of training on evidence impact and the link between rational
belief and desirable belief. Consequently, after the students had been
tested on the critical material, a somewhat smaller sample was pre-
sented with experimental messages and measurement items identical to

those used in Phase I. These constituted the Phase II manipulations.

fiatioaale for Hypotheses

The conceptualization of the relation of evidence to belief devel-
oped in this chapter represents a rather radical departure from the
more simplistic notions of previous evidence researchers. This study
and its immediate predecessor (Cushman and Smith, In Preparation) were
therefore designed primarily to provide descriptive bases for an
ongoing research endeavor rather than to test specific deductions from
the theory of evidence being developed. Nevertheless, the design of
the study does permit tests of a number of preliminary hypotheses.
These hypotheses and their derivations are outlined below.

The most basic issue in evidence research is the question of how
much impact various forms of evidence have on belief. In our study,
nine combinations of attack and constructive evidence-based arguments
were developed and their effects on rational and desirable belief were

measured. 0n the basis of the model presented in Figure 2 and a number

26
of theoretically based assumptions (which are reported below) we will
predict rank orders of rational belief and desirable belief scores
across all nine treatments. Separate predictions will be made for
Phase I and Phase II. Before detailing these predictions, however,
several preliminary points should be noted. First, on the basis of
pretest results, it was assumed that subjects would load quite heavily
on the attack side of the proposition regardless of the evidence treat-
ment employed. As a result, the great majority of our predictions will
be phrased in terms of a decrease in rational belief and desirable
belief. Second, all desirability arguments were generated by pretest
subjects drawn from the same general population as the subjects in the
larger study. Desirability evidence was then devised to support these
arguments. This insures that, in general, desirability evidence will
link with the value systems (metaphysical dreams) of message recipients
on the same side of the proposition as the evidence. Third, all
belief predictions will be phrased in terms of the cell numbers
assigned in Figure 3 above. The addition of the letter "A" to the
number designates the appropriate cell in Phase II.

Our first hypothesis relates to the effects of evidence on
rational belief for Phase I subjects. Our predictions are derived
from the model presented in Figure 2 and the following five assump-
tions. Where appropriate, the source or sources for the assumption
will be listed parenthetically.

1. Treatment 1 (No Evidence versus No Evidence) will provide
baseline ratings of rational belief on the proposition.

2. The link between rational and desirable belief is non-recip-
rocal (Weaver, 1948; McGuire, 1960a,b; Dillehay, Insko and
Smith, 1966) .

 

 

27

3. Desirability evidence that fails to link with the value
system of the message recipient will have the same
effect on rational belief as no evidence (Weaver, 1948;
Cushman and Smith, In Preparation).

4. In general, supportive desirability evidence will have
a stronger impact on rational belief than truth evidence
(Weaver, 1948, 1963; Filion, 1972; Cushman and Smith, In
Preparation).

5. Truth evidence, in the absence of desirability evidence,
will have an impact on rational belief regardless of
the initial position of the message recipient.
0n the basis of these assumptions and the general model we have derived
the following major hypothesis:

H1: Posttest ratings of rational belief for phase I
subjects will display the following pattern:
7) 1,4,9) 3,6) 8) 2, 5.

Our next hypothesis relates to the effects of evidence on desirable
belief. Once again, our predictions are based on the model presented
in Figure 2 and the following three assumptions (cf., Assumptions 1,
2 and 3 above).

1. Treatment 1 will provide baseline ratings of desirable
belief on the proposition.

2. The link between rational belief and desirable belief
is nonpreciprocal.

3. Desirability evidence, if it fails to link with the value
system of the message recipient, will have the same effect
on desirable belief as no evidence.
0n the basis of these assumptions and the general model we have derived
the following hypothesis:
H2: Posttest ratings of desirable belief for Phase I sub-
jects will be higher in treatments 1, 3, 4, 6, 7 and 9
than in treatments 2, 5 and 8.
In addition to investigating the effects of different forms of

evidence on different forms of belief, the current study was also

 

28
intended to discover the effects of certain variables on the link
between rational belief and desirable belief. This approach was sug-
gested by the finding of Dillehay, Insko and Smith (1966) that the link
between truth and desirability ratings in one of their experiments vir-
tually disappeared when upper level college undergraduates were used as
subjects, although it was quite prominent when lower level undergrad-
uates were used. The authors argued, somewhat nebulously, that either
intelligence or education was the critical intervening variable. After
a careful review of the Dillehay, Insko and Smith study and its prede-
cessor, McGuire (1960b), we have rejected the former possibility while
retaining the latter as a working hypothesis. Our reasons for reject-
ing intelligence as the intervening variable are twofold. First, it
seems unlikely that the difference in intelligence between upper and
lower level undergraduates is sufficiently large to produce the
observed differences in correlations of truth and desirability ratings.
Second, even though McGuire used high school seniors and college fresh-
men with a modal high school grade average below the 30th percentile
for their state, and Dillehay, Insko and Smith used a more represene
tative sample of lower level undergraduates, the expected link between
truth and desirability ratings was found in both studies. The link
largely disappeared only when Dillehay, Insko and Smith moved to upper
level undergraduates. We have therefore decided to test only the
effect of education level on the link between rational belief and
desirable belief. Before articulating our hypothesis, however, we
must add one theoretical refinement. According to Weaver's theory,

the influence of a man's metaphysical dream on his beliefs about

 

29
reality is both pervasive and permanent. Thus, while it might be pos-
sible to suppress the link between rational belief and desirable belief,
we would never expect the link to disappear entirely. This contention
receives some support from the Dillehay, Insko and Smith study. They
found that the correlation between truth and desirability ratings for

their upper level undergraduates did not disappear entirely but was

 

instead reduced to the non-significant (for their sample) value of .21
(1966, p.651). On the basis of these arguments, then, we may now pro-
pose the following hypothesis:
H3: Across all treatments, for both Phase I and Phase II
subjects, the correlation between rational belief
and desirable belief ratings will be lower for upper
level undergraduates (i.e., Juniors and Seniors) than
for lower level undergraduates (i.e., Freshmen and
Sophomores), but will always be significantly greater
than zero.

Confirmation of the foregoing hypothesis would have a profound
influence on future theories of evidence, but it would do little to
explain the phenomenon in question. Simply knowing that more highly
educated people are better able to separate rational belief from desir-
able belief tells us nothing about how they acquired that capability.
One major possibility is that the average upper level undergraduate
has had at least some formal training in reasoning, while the average
lower level undergraduate has not. In order to test this possibility,
subjects were drawn from the same population both before (Phase I) and
after (Phase II) receiving training in argumentation and the proper use
of evidence. 0n the assumption that this training indirectly contribu-
ted to the subjects' ability to separate rational belief from desirable

belief (subject of course to the theoretical limitation noted above)

30
we have derived the following exploratory hypothesis:

H4: Across all treatments, the correlation between rational
belief and desirable belief will be lower for Phase II
subjects than for Phase I subjects, but in both cases
will be significantly greater than zero.

Our final two hypotheses relate to the effects of evidence on
rational and desirable belief for Phase II subjects. Unfortunately,
the hypothesized suppression of the link between rational and desirable
belief for these subjects makes predictions as to the effects of desir-
ability evidence on rational belief problematic. Depending on the mag-
nitude of this suppression, desirability evidence could be more power-
ful, less powerful, or approximately as powerful as truth evidence in
its effects on rational belief. Solely on the basis of the rather
meager evidence provided by the Dillehay, Insko and Smith (1966) study,
we have chosen the last alternative in phrasing our hypothesis. Thus
while Assumptions 1, 2, 3 and 5 are still applicable, Assumption 4 must
be changed as follows:

4a: Supportive desirability evidence will have approximately
the same impact on rational belief as truth evidence.

0n the basis of these five assumptions and the general model, we may
now derive the following hypothesis:

H5: Posttest ratings of rational belief for Phase II sub-

jects will display the following pattern:
7A) 1A, 4A, 8A, 9A) 2A, 3A, 5A, 6A.

Although the hypothesized suppression of the link between rational
belief and desirable belief should have a significant effect on the
relation of desirability evidence on rational belief, it should have
no effect on the relation between evidence and desirable belief.

Therefore, on the basis of the three assumptions reported for

31
Hypothesis 2 above and the general model, we may derive the following
final hypothesis:
H6: Posttest ratings of desirable belief for Phase II

subjects will be higher in treatments 1A, 3A, 4A,
6A, 7A and 9A than in treatments 2A, 5A and 8A.

 

CHAPTER II

METHODOLOGY

Pretest

0n the twin bases of the experience gained in conducting the first
experiment in this series (Cushman and Smith, In Preparation) and the
theoretical position developed in this study, three criteria were
derived to guide the construction of theoretically and empirically
adequate message treatments. First, it was decided that the experimen-
tal proposition should relate to an issue of high interest and impor-
tance to the experimental audience. This would not only insure a
closer correspondence between the experimental situation and the idela
rhetorical situation (Bitzer, 1967) which it should approximate, but
would also increase the pragmatic utility of the findings. Second,
the experimental proposition should be one on which the audience loaded
heavily on either the positive or negative side. This would permit
generation of a single set of belief predictions and would eliminate
the problem of posttest division into similar belief groups. On the
other hand, it was recognized that with such a proposition any exper-
imentally induced changes in belief on the proposition would tend to
be both small and in the same direction as original belief. Third,
any desirability evidence and arguments would have to link with the
general value systems of at least a majority of the experimental audi-
ence. As we argued in Chapter I, this is a necessary condition for

32

 

 

33
the effective use of desirability evidence in any kind of rhetorical
situation.

In order to gather the information necessary to meet these three
criteria, a pretest was conducted in January, 1972, approximately four
months before the beginning of Phase I data collection. Subjects in
the pretest were 32 undergraduate students enrolled in one section of
the Communication 100 course at Michigan State University. The sub-
jects were presented with a seven-page booklet (see Appendix A) which 1
contained five propositions and scales for rating the truth, desirabi- A
lity and importance of each proposition. In addition to completing

these scales, each subject was also asked to list one major reason why

 

he or she felt each proposition was desirable, and one major reason
why he or she felt the proposition was undesirable. On the basis of
the information gathered in the pretest, a single proposition ("Uni-
versity administrators must be given the right to search student dormi-
tory rooms for drugs.") was chosen for use in the larger study. This
proposition received a mean pretest truth rating of 3.11 on a tenspoint
scale ranging from "0 - 10% true" to "91 - 100% true", a mean pretest
desirability rating of 1.68 on a four-point scale ranging from "Very
undesirable" to "Very desirable", and a mean pretest importance rating
of 4.36 on a five-point scale ranging from "Very unimportant" to "Very
important". Desirability evidence and arguments on the proposition
were generated on the basis of the responses to the two openpended
questions. In the pretest, 91% of the subjects listed "loss of free-
dom" or "violation of individual rights" as the most undesirable fea-

ture of the proposition, while 56% listed "rehabilitation of drug users"

34
as the most desirable feature. The lower percentage for the desirable
feature reflects both a broader range of responses and the inability of

some subjects to foresee any desirable consequences to the proposition.

Samples
The subjects for Phase I of the study were 273 undergraduate stu-

dents enrolled in eleven sections of the Communication 100 course at
Michigan State University. The sample was about evenly divided between
males and females (50.19% versus 49.81%). All questionnaires were
usable although eleven subjects failed to answer one or more demogra-
phic items. Phase I data collection took place from.May 8 to May 10,
1972.

The subjects for Phase II of the study were 166 undergraduate stu-
dents enrolled in seven of the same eleven sections of the Communication
' 100 course as in Phase I. It was assumed that most if not all of the
Phase II subjects had also participated in Phase I of the study. Once
again, the sample was about evenly divided between males and females
(50.63% versus 49.37%). All questionnaires were usable although nine
subjects failed to answer one or more demographic items. Phase II data

collection took place from May 31 to June 2, 1972.

Operationalizatiope of Indepepdent Variables

The primary independent vatiable in both phases of the study was
evidence. Three types of evidence were identified--no evidence, desir-
ability evidence and truth evidence. Each type of evidence was used to

support one constructive argument and one attack argument on the

35

proposition. Thus, a total of six evidence treatments were construc-
ted. Each treatment is outlined below (see also Appendix B).

1. No Evidence - Constructive Argument: This treatment consisted
of a 71-word paragraph containing an introductory sentence,
two one-sentence authority-based definitions of terms appear-
ing in the proposition, and a concluding sentence calling for
support of the proposition.

2. No Evidence - Attack Argument: This treatment consisted of a
73-word paragraph containing an introductory sentence, two
one-sentence authority-based definitions of terms appearing
in the proposition, and a concluding sentence calling for
opposition to the proposition.

3. Desirability Evidence - Constructive Argument: This treatment
consisted of a 79-word paragraph containing an introductory
sentence, two one-sentence authority-based opinion statements
testifying to the desirability of the proposition, and a con-
cluding sentence calling for support of the proposition.

4. Desirability Evidence - Attack Argument: This treatment con-
sisted of an 80—word paragraph containing an introductory sen-
tence, two one-sentence authorityabased opinion statements
testifying to the undesirability of the proposition, and a
concluding sentence calling for opposition to the proposition.

5. Truth Evidence - Constructive Argument: This treatment con-
sisted of an 84-word paragraph containing an introductory sen-
tence, two one-sentence authority-based statements of fact
supporting the truth of the proposition, and a concluding sen-
tence calling for support of the proposition.

6. Truth Evidence - Attack Argument: This treatment consisted
of a 78-word paragraph containing an introductory sentence,
two one-sentence statements by authorities citing facts to
support the falsity of the proposition, and a concluding sen-
tence calling for opposition to the proposition.

The second independent variable in the study was training in evi-
dence usage and argumentation. This training was conducted as an inte-
gral part of the Communication 100 course structure. The training
sequence was divided into four parts. First, all Communication 100
students were required to read a brief paper outlining the model of an

argument developed by Toulmin (1959) and stressing the proper use of

 

36
evidence in argumentation. Second, each instructor led a 15-20 minute
classroom discussion on the content of the article. Third, all stue
dents were required to participate in the construction of written cases
urging adoption of a specific solution to a problem. These cases were
then critiqued and graded, partially on the basis of how well the stu-
dent had been able to apply the principles of argumentation expounded
in the assigned article. Fourth, student comprehension of the material
on argumentation and evidence usage was tested as part of a broader
course-wide examination. The training sequence was initiated three-to-
five days after completion of Phase I data collection and was termine
ated approximately one week before the beginning of Phase II data col-
lection.

In order to test our third hypothesis, it was necessary to distine
guish between upper and lower level undergraduates. In this study, all
subjects who checked the categories "Freshman" or "Sophomore" in
response to the demographic question "What is your college class?" were
classified as lower level undergraduates. All subjects who checked the
categories "Junior" or "Senior" in response to the same question were
classified as upper level undergraduates. 0f the Phase I sample,
72.93% were classified as lower level undergraduates, while the corres-

ponding figure for the Phase II sample was 68.79%.

Operationalizations of Depandent Variables
The two dependent variables in both phases of the study were

 

rational belief and desirable belief. Our operationalizations of these

variables were based on measurement instruments developed by McGuire

 

 

37

 

Part A: Part B:
(a) 0 - 10% true (a) Very desirable
(b) 11 - 20% true (b) Desirable
(c) 21 - 30% true (c) Undesirable
(d) 31 - 40% true (d) Very undesirable

(e) 41 - 50% true
(f)_____51 — 60% true
(g) 61 — 70% true
(h) 71 - 80% true
(i) 81 - 90% true
(3).- 91 -100% true

 

Figure 4. Dependent measurement instruments

 

(1960b) to measure judgements of the truth and desirability of syllo-
gistic premises and conclusions. In our study, rational belief was
operationally defined as the response to the question "How true is the
proposition that University administrators must be given the right to
search student dormitory rooms for drugs?" measured on the scale shown
in Part A of Figure 4. Subjects were instructed to place an "X" in
the blank that most accurately described their evaluation of the truth
of the proposition. Responses were scored from 1 to 10, with 10 repre-
senting a judgement of 91 - 100% true.

Desirable belief was operationally defined as the response to the
question "How desirable is it that University administrators be given
the right to search student dormitory rooms for drugs?" measured on the
scale shown in Part B of Figure 4. Subjects were instructed to place
an "X" in the blank that most accurately described their estimate of
the desirability of the proposition. Responses were scored from 1 to 4

with 4 representing a judgement of Very desirable.

 

, rum: K.

 

38

Measurement instruments quite similar to those shown above were
used in studies by McGuire (1960a,b) and Dillehay, Insko and Smith
(1966); instruments identical to those shown above were used in the
first study in this series (Cushman and Smith, In Preparation). Before
adopting these instruments for use in our evidence studies, however,
several checks of their face validity were accomplished. These checks
were conducted as part of a pretest for the Cushman and Smith study.

In this pretest, 36 high school students were interviewed in depth to

f},

determine (1) how they normally conceptualized truth and desirability,

and (2) whether they normally applied these categories in evaluating

 

propositions similar to the one employed in this study. Results of
the pretest indicated that these subjects (1) generally conceptualized
truth in probabilistic terms, (2) generally conceptualized desirabi-
lity in terms of a small number of broad verbal categories, and (3)
normally applied these categories (i.e., truth and desirability) in
evaluating diverse propositions. It was assumed in this study that

these findings could be generalized to a college student population.

Questionnaire Copstructiop

Each subject in both phases of the study received a four-page
booklet containing instructions, evidence treatments and dependent
measures. There were a total of nine versions of the basic booklet,
which differed only on the combination of attack and constructive argu-
ments included (e.g., Constructive Argument - No Evidence versus
Attack Argument - No Evidence, Constructive Argument - No Evidence ver-

sus Attack Argument - Desirability Evidence, and so on). Each version

39

of the booklet therefore corresponded to one of the nine basic treat-
ment groups in either phase of the study. The first page of each book-
let contained general instructions, an example of the type of proposi-
tion used in the study, and specific instructions on how to complete
the dependent measures of rational belief and desirable belief (see
Appendix B for full versions of all booklets). The second page of each
booklet contained a statement of the proposition, a one paragraph con-
structive argument on the proposition, a one-paragraph attack argument #11
on the proposition, and four completion questions about the substantive

content of each of the pieces of evidence used in the arguments. The

 

constructive argument preceded the attack argument in all nine versions
of the booklet. The third page of each booklet contained the dependent
measures of rational belief and desirable belief and a request for ver-
bal justification of the subject's judgements. The fourth and final

page of each booklet contained eight demographic questions.

Data Collectiongrocedurea

The sequence of data collection procedures for Phase I of the
study was as follows. All copies of the blank questionnaires were
assembled as a single set, and the order of appearance of the nine dif-
ferent treatments was randomized within the set. Treatments 1, 5 and 9
(see p. 25 above) were overrepresented within the set to provide addi-
tional information on these critical cells. The randomized set of
questionnaires was then divided into eleven subsets of approximately
equal size. Each of the eleven instructors of daytime Communication

100 sections for Spring term, 1972, was given one subset of booklets

40

for administration to his class. All instructors were asked to: (1)
distribute the questionnaire booklets approximately fifteen minutes
before the end of the scheduled class period, (2) introduce the ques-
tionnaire as part of an opinion survey being conducted by the Depart-
ment of Communication, and (3) answer any questions about the question-
naire with the statement "I don't know".

Data collection procedures for Phase II of the study were identi-
cal to those for Phase I with two exceptions. First, treatments 1, 5
and 9 were not overrepresented in the second set of questionnaires.
Second, questionnaires were distributed in only seven of the original
eleven Communication 100 sections. This occurred because four of the
instructors were behind schedule (Phase II data collection took place
during the last week of classes for Spring term)and could not allot
class time for questionnaire administration. It could be argued that
the seven instructors who were able to distribute the Phase II ques-
tionnaires were more efficient than the four who could not, and were
therefore more effective teachers as well. If this were the case, we
would expect the hypothesized impact of training on the relation
between rational belief and desirable belief (Hypothesis 4) to be
greater than if all eleven instructors had participated in Phase II
data collection. This would tend to increase the probability of con-

firming Hypothesis 4.

Control Procedures
In Chapter I we identified several uncontrolled intervening varia-

bles which may have confounded the results of previous studies of the

 

41
effects of evidence. With one exception, each of these variables was
explicitly controlled in both phases of this study. The first varia-
ble, source credibility, was controlled by using written anonymous mes-
sages. The only attribution of the messages was to the Department of
Communication, which we assumed would be perceived as moderately high
in credibility. On the basis of McCroskey's (1969) findings, we would
therefore expect source credibility to depress the effects of evidence
in all treatment groups, if it had any effect at all. This would make
it more difficult for us to confirm Hypotheses 1, 2, 5 and 6. The
second variable, audience familiarity with the evidential data, was
controlled in Phase I by creating all the evidence used in the experi-
mental messages. Thus all subjects in Phase I received equally new and
unfamiliar evidence. Unfortunately, it was not possible to completely
control for familiarity in Phase II of the study. The same messages
used in Phase I were randomly distributed to subjects in Phase II, the
vast majority of whom.had participated in Phase I. As a result, we
must assume that a substantial minority of Phase II subjects were famil-
iar with one or both of the messages they received. The third variable,
redundancy, was controlled by simply constructing nonpredundant mes-
sages. This was facilitated in the two "no evidence" messages by our
use of authority-based definitions in place of evidence. In addition,
an attempt was made to control for message length. The six messages
used in the study varied from 71 to 84 words, with a mean length of 78
words. A fourth control problem.relates to the structure of evidential
statements. As we pointed out in Chapter I, each piece of evidence

used in previous evidence studies generally contained three separate

 

Table 2. Evidence sources and their qualifications

 

Apgument Source Qaalificatioge

Constructive Dr. Joseph Elder Professor of Sociology, University
of Wisconsin

Constructive Peter Leonard Associate Director, National Edu-
cation Association

Attack Dr. Joseph Woelfel Professor of Sociology, University
of Illinois

Attack James Small Research Assistant, American Asso-
ciation of University Professors

 

and possibly interacting sources of persuasive potential--the eviden-
tial data, the source of the evidential data and the qualifications of
the source of the evidential data. In our study, each piece of evi-
dence was also made up of data, source and qualifications, but sources
and qualifications were held constant across all forms of evidence on
each side of the proposition. These sources and their qualifications
are listed in Table 2.

The four control problems discussed above are common to most
studies of the effects of evidence. In addition to these, three pos-
sible sources of bias were identified that were more or less unique
to the type of experimental procedure followed in our study. First,
there may have been critical differences in the way the eleven instruc-
tors administered the questionnaire booklets. Second, there may have
been critical differences in the way individual instructors presented

the training materials on argumentation and evidence usage. Third,

 

43

the various classroom sections may have differed on one or more criti-
cal population parameters. All of these three possible sources of bias
were controlled in our study by randomizing the experimental treatment

booklets before distributing them to the eleven instructors.

Desigp
The study was in the form of a 3x3, replicated, after-only design.

All experimental treatment groups received some combination of experi-
mental messages, however the No Evidence versus No Evidence cells in
both phases of the study (cells 1 and 1A in Figure 5 below) were treated
as no-manipulation control groups. The design is summarized in Figure 5
below, which also includes data on the total sample sizes and the sizes

of each treatment group.

 

 

Phase I: N = 273

 

 

 

 

 

 

 

 

 

Attack
No Dg. Truth
(1) (2) 3)
No
n= n=2 n=26
(4 5 (65
Constructive Des .
n=21 n= 6 n-
7 (8 , (9
Truth
n=2; n=2} :1:
Phase II: N = 166
Attack
No Des. Truth
<1A) (2A? 3A)
No
n=20 n=19 n=18
III) (a) 6A)
Constructive Des .

Truth

 

 

 

 

 

 

Figure 5. Full experimental design with sample sizes

 

 

CHAPTER III

RESULTS

Effects of Evidence: Phaaa_l

The major hypotheses in this study relate to the effects of evi-
dence on rational belief and desirable belief. Specifically, we pre—
dicted in our first hypothesis that, for subjects in Phase I, posttest
ratings of rational belief would display the following pattern: 7.) 1,
4, 9 ) 3, 6 ) 8 ) 2, 5. The actual pattern of rational belief ratings
for Phase I is shown in Table 3. In this Table, as in Tables 4, 6 and
7 below, we have chosen to describe our findings with two sets of fig—
ures instead of the conventional single set. First, the number in
parentheses in each cell of the treatment matrix indicates the mean
value of rational belief for that cell. Second, the centered number
in each cell indicates the product-moment correlation between a dichot-
omous variable representing that particular treatment and posttest
ratings of rational belief, computed across the entire (N = 273 or 166)
sample. Each of these correlation coefficients therefore indexes the
apparent impact of each treatment relative to all other treatments in
the matrix. The advantage of this technique is that it tends to high-
light extremely small treatment differences, an important consideration
in a primarily descriptive study such as this one. Unless otherwise

indicated, all statistical tests reported in this chapter will be based

45

46

 

Table 3. Rational belief ratings for Phase I subjects

 

 

 

 

 

 

 

 

 

 

Attack
No Desirable Truth
(3.03) (1.41) (2.12)
No
.113. -,157 :,0
(2.90) (1.78) (2.75
Constructive Desirable
.069 -0127 1052
(2.33) (2.13) (3.01.)
Truth
-.008 -.O37 .135
Predicted: 7) 1,4,9) 3,6) 8) 2,5
Discovered: 9)1)4) 6) 7)8)3>5)2

 

on the correlational data rather than the treatment means. The level
of significance used for all analyses was .05.

The most obvious conclusion that may be drawn from Table 3 is that
none of the evidence combinations had a very great impact on rational
belief. This finding, which was not unexpected, will be discussed in
some detail in Chapter IV. Nevertheless, comparison of the "Predicted"
and "Discovered" inequalities reveals that the pattern of belief rat-
ings discovered closely approximates the predicted pattern. This sime
ilarity is reflected in a significant Spearman rank correlation coeffi-
cient of .769. A careful inspection of the data discloses only two
substantial deviations from the predicted pattern. First, the correla-
tion between treatment and rational belief ratings for Treatment 7 is
much smaller than predicted. Second, the correlations for Treatments

3 and 6, which we expected would be of nearly equal magnitude, are

47

 

 

 

 

 

Table 4. Desirable belief ratings for Phase I subjects
Attack
No Desirable Truth
(1.60) (1.17)’ (1.387"
No
2076 "a1 "o035
(1.71) (1.24 (1.38)
Constructive Desirable
.1015 ”914-4 -0038
(1.48) (1.1.2) (1.76)
Truth
.007 -.020 .187

 

 

 

 

 

 

 

actually separated by a fairly large, albeit non-significant, margin.
Bearing these deviations in mind, we must conservatively conclude that
Hypothesis 1 is only partially supported by the experimental data.

Our second hypothesis predicted that posttest ratings of desirable
belief would be higher in treatments 1, 3, 4, 6, 7 and 9 than in treat-
ments 2, 5 and 8. The mean values of desirable belief and the correla-
tions between treatment and desirable belief ratings for Phase I sub-
jects are shown in Table 4. In order to evaluate our hypothesis, the
general prediction statement was broken down into the eighteen two-
element inequalities that it implies (e.g., 1 is greater than 2). Each
of these inequalities was then compared against the experimental data,
using simple numerical magnitude as the criterion for determing the
direction of the actual relationship. Sixteen of the eighteen inequal-
ities were in the direction predicted by Hypothesis 2. The two deviant
cases were 3 and 6 smaller than 8. If we assume that the random. proba-

bility of confirming any one of these eighteen predictions is equal

48
to .50, then the probability of confirming sixteen or more of the
eighteen predictions is less than .001. We may therefore conclude
that Hypothesis 2 is supported by the experimental data. we nust note,
however, that the data suggest a more complex relationship between
evidence and desirability belief than that envisioned in our model.
Specifically, while our model would lead us to expect only small, rane
dom differences in ratings of desirable belief and the corresponding
treatment/belief correlations within the two treatment subsets (i.e.,
1, 3, 4, 6, 7, 9 and 2, 5, 8), our data reveal a number of relatively
large and often significant differences between treatments within the
same subset. Thus, while the predictions derived from the model are
generally accurate, they would also seem to be incomplete. We will

discuss this finding at some length in Chapter IV.

Effects of Education Level_apd Trainipg

In addition to the hypotheses relating to the effects of evidence
on belief, we also formulated two hypotheses about the effects of out-
side variables on the relation between rational belief and desirable
belief. In Hypothesis 3, we predicted that the correlation between
rational belief and desirable belief would be higher for lower level
undergraduates (LLU) than for upper level undergraduates (ULU) in both
; phases of the study. The actual correlations are shown in Table 5. In
both Phase I and Phase II, the predicted relationship is reversed--
rational belief and desirable belief are correlated more highly for
upper level undergraduates than for lower level undergraduates,

although the difference between correlations is non-significant in

49

 

Table 5. Belief correlations

 

 

 

 

 

 

Sample N r Difference p
Phase I - LLU 194 .486

_,169 --_
Phase I - ULU 72 .655
Phase II - LLU 108 .482

-0174 ---
Phase II — ULU 72 .656
Phase I 273 .530

-.031 ---
Phase II 166 .561

 

both cases (using Fisher p to a transformations, the a value for Phase
I is 1.802 and for Phase II is 1.465). We must therefore conclude that -
Hypothesis 3 is not supported by the data.

In Hypothesis 4 we predicted that the correlation between rational
belief and desirable belief would be higher for Phase I subjects than
for Phase II subjects, due to the effects of training in argumentation
and evidence usage. The actual correlations are also shown in Table 5.
Once again, the predicted relationship is reversed--the correlation
between rational belief and desirable belief is slightly higher for
Phase II subjects than for Phase I subjects. We must therefore cone

clude that Hypothesis 4 is not suppOrted by the experimental data.

Effects of Evidence: Phase II
Based on the postulated impact of training on the relation between

rational belief and desirable belief, we fomulated a separate set of

 

 

50

 

Rational belief ratings for Phase II subjects

 

 

 

 

 

 

 

 

 

 

Attack
No Desigable Truta
(2.35) (3.00) (1.89)
No
2.25) (1.71.) (3.11)
Constructive Desirable .
1019 -.105 .118 )1
(2.33) (2.32) (2.37)
Truth
3006 '1010 “3001

Predicted: 7A > 1A, 4A, 8A, 9A) 2A, 3A, 5A, 6A

Discovered: 6A) 2A) 9A) 1A) 7A) 8A> 4A) 3A) 5A

 

predictions in Hypothesis 5 as to the effects of the various evidence
combinations on rational belief for Phase II subjects. Given the appa-
rent failure of the training manipulation, however, we would not expect
to find much confirmation for Hypothesis 5. Such is indeed the case.
As shown in Table 6, there is a wide disparity between the predicted
pattern of results and the pattern discovered. For example, although
we expected only small, random differences between treatments 2A, 3A,
5A and 6A, in reality the correlations for treatments 2A and 6A are

the highest of all the nine treatments, while the correlations for
treatments 3A and 5A are the lowest. We must therefore conclude that
Hypothesis 5 is not supported by the data. Of course, given the fail-
ure of the training manipulation, it could be argued that the appro-
priate set of predictions for Phase II subjects would be that set forms

ulated in Hypothesis 1. Unfortunately, comparison of the treatment

51

 

Table 7. Desirable belief ratings for Phase II subjects

 

 

 

 

 

 

Attack
No Desirable Truth
N (1.55) (1753) (1.33)
o
.072 4057 '90
(1.25) (1.68) 1.39
Constructive Desirable
-°085 O1 -1017 1H
1.28) 1.47 (1.26)
Truth
-.O76 £028 -.086

 

 

 

 

 

 

 

' ‘3.“ ‘-

order predicted by Hypothesis 1 with the discovered order yields a non-
significant Spearman rank order correlation coefficient of .121. Thus
we are forced to reject this alternative hypothesis as well.
A Our last hypothesis, which is identical to Hypothesis 2, predicted
that posttest ratings of desirable belief would be higher in treatments
1A, 3A, 4A, 6A, 7A and 9A than in treatments 2A, 5A and 8A. The actual
means and correlations for Phase II subjects are shown in Table 7. 'As
before, the general prediction statement was broken down into eighteen
two-element inequalities for evaluation. Only two of the eighteen ine-
qualities were in the direction predicted. We must therefore conclude
that Hypothesis 6 is not supported by the data. The probability of
confirming two or less predictions out of eighteen, however, if the
random probability of confirming any single prediction is equal to .50,
is less than .001. Thus we must consider the possibility that the

.observed pattern of results was strongly influenced by one or more

52
uncontrolled intervening variables operating in opposition to the
hypothesized relationships, or superceding them in this instance. We

will discuss this possibility at some length in the following chapter.

 

CHAPTER IV

DISCUSSION

M

One of the most basic tenets of rhetorical theory is that evidence
should have a strong and reliable impact on belief. Unfortunately, the
great majority of empirical studies of the relationship between evi-
dence and belief have produced conflicting or negative results. It was
this problem that gave rise to the present study. We began our inquiry
with a critical review of both rhetorical theory and previous evidence
research. On the basis of this review we identified nine theoretical,
definitional and procedural shortcomings of the earlier research which
may have contributed to the disappointing results of that research. We
then developed a model of the relation of evidence to belief, grounded
in a set of distinctions drawn primarily from the rhetorical theories
of Aristotle and Richard Weaver, which seemed capable of overcoming
those shortcomings. In our research, we tested four hypotheses derived
from.the model as well as two hypotheses derived from.the empirical
studies of McGuire (1960a,b) and Dillehay, Insko and Smith (1966) con-
cerning the effects of certain outside variables on one of the links in
the model. In particular, we sought to determine: (1) the effects of
different forms of evidence on rational and desirable belief; (2) the
effects of training in argumentation and evidence usage on the link

53

 

54
between rational and desirable belief; and (3) the effects of general
education level on the link between rational and desirable belief.

Our research began with the construction of six evidence-based
messages. Three of the messages supported a single experimental propo-
sition, while three opposed the same proposition. One of each type of
message was supported by no evidence statements, one of each was sup-
ported by desirability evidence statements, and one of each was sup—
ported by truth evidence statements. Nine experimental treatments were
then composed using all possible combinations of the two types of mes-
sages. In Phase I of the study, these treatments were randomly dis-
tributed to 273 undergraduate students who read the messages and come
pleted scales measuring judgements of the truth and desirability of the
proposition. Several weeks later, after the students had studied, dis-
cussed and applied material on the proper structure of arguments and
the place of evidence in argumentation, the same nine treatments were
again randomly distributed to 166 undergraduates, most of whom it was
assumed had participated in the Phase I manipulation. A total of six
hypotheses were tested in Phases I and II of the study. The first two
hypotheses predicted rank orders of rational belief and desirable
belief ratings by treatment for the Phase I subjects. Both of these
hypotheses were at least partially supported. The next two hypotheses
predicted the effects of training and education level on the correlation
‘between rational and desirable belief. Neither of these hypotheses was
supported. The last two hypotheses predicted rank orders of rational
belief and desirable belief ratings by treatment for the Phase II sub-

jects. Once again, neither of these two hypotheses was supported. The

55
following two sections of this chapter will be devoted to a more thor-

ough discussion of these results.

Effects of Evidence on Belief: Conclusions

Perhaps the most obvious conclusion that may be drawn from our
results is that none of the evidence combinations had a particularly
marked impact on either rational or desirable belief. This is reflected
in the uniformly low correlations between treatment and belief ratings.
We must emphasize, however, that this finding does not necessarily
imply that evidence is only a minor determinant of belief. By our
analysis, three factors in the experimental situation most probably
acted to suppress the effects of evidence across all treatment groups.
First, the evidence-based messages used in the study were not intended
to be especially powerful manipulations. Each subject was exposed to
a total of only eight sentences of persuasive material, four sentences
on each side of the proposition. Given the magnitude of our manipula-
tion attempts, therefore, we would not expect to find dramatic changes
in belief. Second, the proposition used in the study was one that was
known to be extremely important to the experimental subjects. we would
expect beliefs on this type of proposition to be more resistant to
change than those on less salient topics. Third, the vast majority of
our experimental subjects loaded heavily on the extreme negative side
of the proposition. While this loading permitted us to more clearly
observe the overall pattern of evidence effects, it also insured that
the magnitude of those effects would be minimal. This is because: (a)

extreme beliefs are known to be highly resistant to change, and (b)

 

 

 

56
ceiling effects often preclude the possibility of discovering or even
producing any intensification in the original beliefs. The scope of
the latter problem is indicated by the fact that 38% of the subjects
in the baseline No Evidence versus No Eridence treatment in Phase I
rated papa the truth and desirability of the proposition at the extreme
lower ends of the respective measurement scales. If we assume that
this figure is representative of the pre-manipulation beliefs of the
entire sample, then we must conclude that more than a third of our sub-
jects could not possibly have shown any response to the Attack argu-
ments. In short, when these three factors are considered, the finding
that none of the evidence treatments was particularly effective in
changing beliefs should hardly be either surprising or discouraging.
What is encouraging is that a number of evidence combinations did pro-
duce significant changes in both rational and desirable belief ratings.
For example, posttest ratings of rational belief for treatments 2 and 5
differed significantly in the predicted direction from the baseline
rating in treatment 1 (t = 3.2091 and 2.8064 respectively, p1( .01), as
did posttest ratings of desirable belief for the same treatments (3 =
2.8122 and 2.8481 respectively, p.< .01). Still, our focus in this
study is not on the amount of difference between specific pairs of
treatments, but rather on the overall pattern of results regardless of
the magnitude of the observed differences. It is for this reason that
all of our belief hypotheses were phrased in terms of rank order pre-
dictions rather than difference predictions. 'We will now consider the
implications of the patterns of results discovered in the study.

By far the most important finding of our study was that the

 

57

patterns of mean posttest ratings of rational belief and desirable
belief for Phase I subjects corresponded rather closely to the pat-
terns predicted in Hypotheses 1 and 2. There were, of course, sev-
eral deviations from our predictions, but with one exception these
were of relatively minor importance to the theoretical position taken
in this study. These deviations will be discussed in detail below.
Prior to that, however, we must first consider our total failure to
confirm the belief predictions for Phase II subjects (Hypotheses 5 r1
and 6). There are three plausible explanations for this failure: (1) 5

the training manipulation had strong and unforeseen effects on the

 

relation between evidence and belief; (2) the results accurately
reflect the actual impact of evidence on belief, therefore implying
that our theoretical position is in error; or (3) the results were

1 biased by the introduction of one or more critical uncontrolled vari-
ables. The true explanation for our failure to confirm Hypotheses

5 and 6 can only be determined by future research. Pending the come
pletion of such research, however, we will tentatively accept the third
explanation as the most plausible. This descision.was based largely on
the fact that we were not able to effectively control for audience
familiarity with the evidence in Phase II, even though, as McCroskey
(1967) has demonstrated, this variable can have a fairly strong impact
on the relation between evidence and belief. As a result of this deci—
sion, we will not attempt either to analyze the Phase II results in
depth or evaluate our theoretical position with regard to those results.
We should note, however, that neither the rational nor desirable belief

ratings for Phase II seem to display any meaningful pattern.

58

Unlike the belief patterns for Phase II, those for Phase I corres-
pond quite closely to our predictions. As is usually the case, however,
the data do suggest that the actual relation of evidence to belief is
somewhat more complex than we had originally anticipated. Specifically,
there are three deviations from our predictions which, if substantiated
by future research, would lead us to propose three modifications to our
theoretical assumptions or model. The first anomaly involves the
impact of treatment 7 (Constructive Truth Evidence versus Attack No
Evidence). In Hypothesis 1 we predicted that the mean rating of
rational belief in treatment 7 would exceed the same rating in the
baseline cell, treatment 1. In reality, the mean rating for treatment
7 was substantially, although non—significantly, lower than that for
treatment 1. Further, the mean rating of desirable belief in treatment
7, which we expected would be approximately the same as that for treat-
ment 1, was also substantially lower than that for treatment 1. In
short, it appears that treatment 7 had a "boomerang" effect on both
rational belief and desirable belief. One possible explanation for
this finding is that the subjects in treatment 7, who loaded heavily
on the Attack side of the proposition, felt cheated by having their
views represented by a No Evidence argument while the opposing views
‘were supported by a Truth Evidence argument. They may have then mani-
fested their displeasure in lower ratings of the truth and desirability
of the proposition. This explanation receives some support from.an
analysis of the subjects' one-paragraph justifications of their judge-
ments. Unlike any other treatment, the modal response in treatment 7

was to dismiss the Attack argument as "ridiculous" or "useless" and

 

59
supply a totally new set of arguments in opposition to the proposition.
Regardless, more research is needed both to establish the existence of
the boomerang effect and to discover its underlying causes.

The second unexpected finding in our results was that the nonesup—
portive Constructive Desirability Evidence message, which we had assumed
would have only a negligible impact on belief, actually appeared to have
a moderately strong impact on both rational and desirable belief. This
effect is shown most clearly by a comparison of the rational belief
ratings for treatments 3 and 6. The relationship is far from.perfect,
however, as can be seen by comparing the desirable belief ratings for
the same two treatments. Given that both the Attack and Constructive
desirability arguments were generated by a pretest sample drawn from
the same general population as our experimental sample, this finding
should perhaps not be surprising. The only theoretical requirement for
the effectiveness of a desirability argument is that it must link with
the value system of the message recipient. Once again, however, we
must await the results of future research to determine the generality
of our finding.

The third, and potentially the most important, deviation from.our
position is the finding that the link between rational belief and desir-
able belief may actually be reciprocal rather than non-reciprocal. This
is indicated by the consistent effects of both Attack and Constructive
truth evidence on desirable belief. If the link between rational and
desirable belief were really non-reciprocal, we would of course not
expect truth evidence to have any impact on desirable belief. In addi-

tion, comparison of the rank orders of rational belief and desirable

 

60
belief ratings of Phase I subjects yields a highly significant Spearman
rank order correlation coefficient of .850 (p <f.01). We would not
expect this correlation to be as high as it is unless the link between
the two belief ratings were reciprocal. Unfortunately, this evidence
provides only indirect support for the existence of a reciprocal link.
Direct support can come only from future research designed specifically
to test for such a relationship. A technique developed by Blalock
(1969), which uses simultaneous differential equations to test for
mutual causation loops among endogenous variables in a path analytic
model, could provide the methodological basis for this research.

Aside from these deviations, the results from.Phase I of the study
provide rather strong empirical support for our theoretical formulation
of the relation between evidence and belief and, more broadly, for the
potential importance of evidence as a variable in any future theories
of persuasion. Practically, the most important finding of our study is
that truth and desirability evidence have a differential impact on
belief, depending upon the initial position of the message recipient.
For our sample, the most effective type of supportive evidence was
desirability evidence. With one exception (treatment 7), the most
effective type of nonesupportive evidence was truth evidence. This
finding alone, if replicated, would represent a significant increment

in our knowledge of the process of persuasion.

Effects of Outside Variables: Conclusioaa
In our third hypothesis we predicted that as the education level

of our subjects increased, the correlation between judgements of the

A

61
truth and desirability of the proposition would decrease. Not only did
our results fail to provide any support for this hypothesis, they actu-
ally suggested that the reverse relationship might be true. For both
our Phase I and our Phase II subjects, as education level increased the
correlation between rational and desirable belief ratings increased as
well, although in both samples the increase was just short of being sta-
tistically significant. This finding could reflect an increased desire
for cognitive consistency among the more highly educated students, a
decreased ability to separate the truth and desirability of a proposi-
tion, or some other factor. Regardless of the cause, however, substane
tiation of this finding would have profound theoretical and social
implications. For this reason, we feel that future research in this
area is imperative.

In Hypothesis 4, our only exploratory hypothesis, we predicted
that training in argumentation and evidence usage would suppress the
link between rational and desirable belief. Our results failed to show
any effects due to the training variable. This failure could reflect
an inadequate operationalization of training, insufficient control of I
audience familiarity with the evidence, inability of the subjects to
apply the training in a realistic judgement situation, rejection of the
training, or the effects of some other variable or combination of vari-
ables. Even given our failure to confirm the hypothesis, however, we
feel that this approach is potentially fruitful. Thus, at least one
more study, preferably using a stronger training manipulation and more

adequate control procedures, should be conducted to test the hypothesis.

 

62

Research Iaplicatiopa

Throughout this chapter we have emphasized the need for extensive
additional research in order to clarify and replicate our findings.
Much of this research will be accomplished in future studies in the
present series. Beyond this, we believe that our findings are strong
enough to justify calling for new research in the following five areas.
First, largely descriptive studies similar to this one should be carried
out using subjects other than high school or college students, in order
to discover whether the effects of evidence vary by population. The
need for this kind of research is underscored by the results of early
comparisons of the belief data collected in this study with similar
data collected in the Cushman and Smith (In Preparation) study. These
comparisons indicate that high school and college students may differ
in the type of evidence they find most convincing in a given situation.
Second, research should be conducted to quantify the effects of the
various types of evidence on rational and desirable belief across a
number of topics and, if necessary, population groups. One of the key
questions in this research should be whether different categories of
topics demand the use of different types of evidence. Third, much of
the earlier evidence research should be repeated, using the operational
definitions of evidence and belief developed in this study, to deter-
mine the effects of variations in the amount and quality of the two
types of evidence. Fourth, and only after the precise relationship
between evidence and belief has been established, additional variables
such as source credibility, audience knowledge of the topic, familiar-

ity of the evidence, and delivery characteristics should be added to

 

41-

63
the basic model and their effects tested. In this way, the present
theory of evidence can be expanded into a comprehensive theory of
persuasion. Fifth, but certainly not least important, a massive
research effort is needed to determine the nature of the relations
between rational and desirable belief and action, and the influence

of evidence on these relations.

.I-l
..A

 

m
m

REFERENCES

Aristotle. De Anima. In R. McKeon (Ed.), The Basic Works of Aristotle.
New York: Random House, 1941.

. Ethica Nicomachea. In R. MbKeon (Ed. ), The Basic works of
Aristotle. New York: Random.House, 1941.

. Rhetorica. In R. McKeon (Ed.), The Basic works of Aristotle.
New York: Random House, 1941.

Becker, S.L. "Research on Logical and Emotional Proofs." Southern y333 M33
Spaech Journal, 1963, XXVIII, pp. 198 - 207. I, 3"

‘. {4‘ d-
J

Bettinghaus, E.P. "The Relative Effect of the Use of Testimony in a

Persuasive Speech Upon the Attitudes of Listeners." Unpublished
master's thesis, Bradley University, 1953.

. Message Preparation: The Nature of Proof. Indianopolis:
Bobbs-Merrill, 1966.

Bitzer, L.F. "The Rhetorical Situation." Paper presented before the
Central States Speech Association, April, 1967.

Blalock, H.M;, Jr. Theory Copptruction: Froa.Verbal to Mathematical
Formulationa. Englewood Cliffs: Prentice-Hall, 1969.

Bostrom, R.N. and R.K. Tucker. "Evidence, Personality and Attitude N 1:33—3
Change." S ech Mono hs, 1969, XXXVI, pp. 22 - 2'7.

~Cathcart, R.S. "An Experimental Study of the Effectiveness of Four
Methods of Presenting Evidence." SEech Monogpams, 1955,

'“‘ Clark, R.A. and G. Hynes. "The Relationship of Judgements of Desirabi-

lity and Likelihood." S ech ano he, 1970, XXXVII, pp. 199 -
206.

Costley, D.L. "An Experimental Study of the Effectiveness of Quantita-
tive Evidence on Speeches of Advocacy." Unpublished master's
thesis, University of Oklahoma, 1958.

Cronkhite, G.L. "The Relation of Scholastic Aptitude to Logical Argument
and Emotional Appeal." Unpublished master's thesis, Illinois
State University, 1961.

64

 

65

. Cronkhite, G. L "Logic, Emotion, and the Paradigm of Persuasion n .3 )1 (3 311
wwech,1964,p, pp, 13 -13 ~' . H -3.

 

Cushman, D.P. "Aristotle's Functional Approach to Emotional Proof."
Unpublished paper, Michigan State University, 1968.

Cushman, D.P. and G. Hauser. "Richard Weaver on Determining When a
Culture is Worth Saving." In Press.

Cushman, D.P. and T.J. Smith, III. "The Relation of Evidence to Belief:
A Test of a New Model." In Preparation.

Dillehay, R.C., C.A, Insko and M.B. Smith. "Logical Consistency and ..
Attitude Change." Journa_1 of Personalit and Social Ps cholo f
1966, m, pp. 646 - 654.

Dresser, W. "Effectiveness of Satisfactory and Unsatisfactory Evidence

in a Speech of Advocacy." Sgech lbnogafls, 1963, XXII,
pp. 302 - 3060

 

,f‘ 'i'$--—_

Filion, S.L. "The Effects of Task and Source Credibility on Evidence
Usage." Unpublished master's thesis, Michigan State University,
1972.

Gardner, J. "An Experimental Study of the Use of Selected Forms of
Evidence in Effecting Attitude Change." Unpublished master's
thesis, University of Nebraska, 1966.

Gilkinson, H., S.F. Paulson and D.E. Sikkink. "Effects of Order and
Authority in an Argumentative Speech." Quarterly Journal of
Spaech, 1954, 3;, pp. 183 — 192.

Grimaldi, W. "Rhetoric and the Philosophy of Aristotle." Classical
Journal, 1958, LIII, pp. 342 - 369.

Harte, T.B. "Audience Ability to Apply Tests of Evidence." Journal of {5"}; "it.
the Aperican Forepaic Associatiop, 1971, VIII, pp. 109 - 115. ' ‘3 A» '
Infante, D.A. "Predicting Attitude from Desirability and likelihood
Ratings of Rhetorical Propositions." Sgech Lbnoggm, 1971 ,
XXXVIII, pp. 321 - 326.

I Kline, J .A. "Interaction of Evidence and Readers' Intelligence on the

Effects of Short Messages." Quarterly Journal of Spaech,1969,
_L_v_, pp. 407 " 4130

Kruger, A.N. "The Ethics of Persuasion: A Re—examination." The Sfiech
Teacher, 1967, pp. 243 - 248.

66

McCroskey, J.C. "Experimental Studies of the Effects of Ethos and
Evidence in Persuasive Communication." Unpublished doctoral
dissertation, Pennsylvania State University, 1966.

. Studies of the Effect of Evidence in Persuasive Communicatiaa.
East Lansing: Speech Communication Laboratory Report 4-67, 1967.

"A Summary of Experimental Research on the Effects of Evidence
in Persuasive Communication." Qaarterly Journal of Sagach, 1969,
LE, pp. 169 - 176.

. "The Effects of Evidence as an Inhibitor of Counter-Persuasion."
Speech Monmm, 1970, XXXVII, pp. 188 — 194.

MCGuire, W.J. "Cognitive Consistency and Attitude Change." Journal of
Abnormal aad Social Psychology, 1960a, Lg, pp. 345 - 353.

. "Direct and Indirect Effects of Dissonance Producing Messages."
Journal of Abnormal and Social Psychology, 1960b, Lg, pp. 354 - 358.

Miller, G.R. "Studies on the Use of Fear Appeals: A Summary and Analysis."
Central States Speech Journal, 1963, XIX, pp. 117 - 124.

. "Evidence and Argument." In G.R. Miller and T.R. Nilsen (Eds.),
Perspectives on_Argamentation. Chicago: Scott, Foresman, 1966.

Miller, G.R. and MLA. Hewgill. "Some Recent Research on Fear-arousing
Message Appeals." Speech Mbnoggapaa, 1966, XXXIII, pp. 377 - 391.

Ostermeier, T.H. "Effects of Type and Frequency of Reference Upon Per-
ceived Source Credibility and Attitude Change." S ech Mono a he,
1967, XXXIV, 137 - 144.

Osgood, C.E., G.J. Suci and P.H. Tannenbaum. The Measurement of Meaniag.
Urbana: University of Illinois Press, 1957.

Ruechelle, R.C. "An Experimental Study of Audience Recognition of
Emotional and Intellectual Appeals in Persuasion." Spaech
Monographs, 1958, XXV, pp. 49 - 58.

 

Sikkink, D.E. "An Experimental Study of the Effects on the Listener
of Anti-Climax Order and Authority in an Argumentative Speech."
Southern Speech Jouraal, 1956, XXII, pp. 73 - 78.

Toulmin, S. The Uses of Argaaent. Cambridge: Cambridge University
Press, 1959.

Troldahl, V.C., D. Costello and G. Robeck. "Strategies for Developing
Messages that Must Compete with Antagonistic Messages." Research
Report, Department of Communication, Michigan State University,

1969.

 

KWIJI' 1LT;

67

Wagner, G.A. "An Experimental Study of the Relative Effectiveness of
Varying Amounts of Evidence in a Persuasive Communication."
Unpublished master's thesis, Mississippi Southern University,
1958.

Warren, I.D. "The Effect of Credibility in Sources of Testimony on
Audience Attitudes Toward Speaker and Message." Spaech
Mono a he, 1971, XXXVIII, pp. 456 -458.

Weaver, R. Ideas Have Conseguences. Chicago: University of Chicago
Press, 1948.

. "Language is Sermonic" in R.E. Nebergall (Ed.), Dimensions
of Rhetorical Scholarship. Norman: University of Oklahoma
Press, 1963.

Whitehead, J.S., Jr. "Effects of Authority-Based Assertion on Attitude
and Credibility." S ech Mbno a be, 1971, XXXVIII, pp. 311 -
315.

Winans, J.A. Public Speakiag. New York: Century, 1915.

Woolbert, C. "Conviction and Persuasion: Some Consideration of Theory."
Quarterly Journal of Pablic Speakiag, 1917, III, pp. 249 - 264.

Yost, M. "Argument from the Point-of-View of Sociology." ter
Jourgal of Public S aki , 1917, III, pp. 109 - 127.

 

APPENDICES

 

‘/

 

APPENDIX A

PRETEST QUESTIONNAIRE

 

68
OPINION PROFILE
INSTRUCTIONS

The purpose of this profile is to obtain your opinions regarding
the probable truth and desirability of several propositions. On each
of the following pages you will find a proposition followed by a series
of questions. For example, here is a proposition similar to the one you
will see in the questionnaire.

SAMPLE PROPETION

Detergents are polluting our nation's rivers.

QUESTIONS

The first question after each proposition will ask you to rate the
probable truth of the proposition on a scale ranging from.0% to 100%
true. Please place an "X" in the blank that most accurately describes
your evaluation of the truth of each proposition.

The second question will ask you to rate the desirability of each
proposition on a scale ranging from "Very Desirable" to "Very Undesirable".
Once again, please place an "X" in the blank that most accurately
describes your estimate of the desirability of each proposition.

The last question on each page will ask you to writeone major reason
why you think the proposition is desirable and one major reason why you
think the proposition is undesirable. Even if you feel very strongly
one way or the other about the proposition, please try to answer both parts
of each of these questions.

ORDER

Please complete all of one page before proceeding to the next page.
Once you have completed a page please do not return to it. You may,
however, return to this page at any time in order to clarify any instruc-
tions that are unclear to you.

 

69

Proposition - University administrators must be given the right to search
student dormitory rooms for drugs.

1. How true do you believe this proposition is?

(a) O - 10% true
(b) 11 - 20% true
(c) 21 - 30% true
(d) :31 - 40% true
(9) 41 _- 50% true
(f) 51 - 60% true
(g) 61 - 70% true
(h) 71 - 80% true
(1) 81 - 90% true
(.1). 91 400% true

2. How desirable do you believe this proposition is?

(a) Very desirable
(b) Desirable

(c) Undesirable
(d) Very undesirable

3. Please write the major reason why you believe this proposition is
desirable and the major reason why you believe this proposition is
undesirable.

(a) I believe this proposition is desirable because:

(b) I believe this proposition is undesirable because:

7O

Propositioa - The University must be given the right to expel aw
student convicted of a crime committed during a violent demonstration.

1. How true do you believe this proposition is?

(a) 0 - 10% true
(b) 11 - 20% true
(c) 21 - 30% true
(d) 31 - 1.0% true
(6). .. 1.1 - 50% true
(f) 51 - 60% true
(g) 61 - 70% true
(h) 71 80% true

(1)—81 - 90% true k...
(.1) 91 400% true

 

2. How desirable do you believe this proposition is?

_tm

 

(a) Very desirable
(b) Desirable ‘\
(c) Undesirable ‘

(d) Very undesirable

3. Please write the major reason.why you believe this proposition is
desirable and the major reason why you believe this proposition is
undesirable.

(a) I believe this proposition is desirable because:

(b) I believe this proposition is undesirable because:

71

Propgsition - The University must take a strong institutional stand
against American participation in the Viet Nam.war.

1.

How true do you believe this proposition is?

(a) o .. 10% true
(b) 11 - 20% true
(c) 21 - 30% true
((1)—31 - 1.0% true
(e) 41 - 50% true
(1‘) 51 — 60% true
(g) 61 - 70% true
(h) 71 - 80% true
(i) 81 - 90% true
(.1) 91 400% true

 

How desirable do you believe this proposition is?

(a) Very desirable
(b) Desirable
(c) Undesirable

(d) Very undesirable

Please write the major reason why you believe this proposition is
desirable and the major reason why you believe this proposition is
undesirable.

(a) I believe this proposition is desirable because:

(b) I believe this proposition is undesirable because:

 

72

Proposition - University police must be prohibited from carrying
firearms on campus.

1. How true do you believe this proposition is?

(a) 10% true
(b) 11 - 20% true
(C)_____21 - 30% true
(d) 31 - 40% true
(6) 1.1 - 50% true
(f) 51 - 60% true
(g) 61 - 70% true
(h) 71 - 80% true
(i) 81 - 90% true
(J) 91 400% true

2. How desirable do you believe this proposition is?

(a) _Very desirable

(b b) _Desirable -.
(c )— _Undesirable “t
(d ) :Very undesirable

 

3. Please write the major reason why you believe this proposition is
desirable and the major reason why you believe this proposition is
undesirable.

(a) I believe this proposition is desirable because:

(b) I believe this proposition is undesirable because:

73

Prgpositioa - The university should prohibit all military recruiting
on campus.

1. How true do you believe this proposition is?

(a) O - 10% true
(b)__1o - 20% true
(c) 21 — 30% true

(d) :31 - 40% true
(e) 41 - 50% true
(f) 51 - 60% true
70% true
(h) 71 - 80% true
(i) 81 - 90% true
(.1)_ 91 -100% true

A
on
v
0\
_8
I

2. How desirable do you believe this proposition is?

(a) Very desirable
(b) Desirable

(c) Undesirable
(d) Very Undesirable

3. Please write the major reason why you believe this proposition is
desirable and the major reason why you believe this proposition is
undesirable.

(a) I believe this proposition is desirable because:

(b) I believe this proposition is undesirable because:

 

 

./ “V

5.

74

How impartant is the proposition "University administrators must be
given the right to search student dormitory rooms for drugs"?

(a) Very important
(b) Important

c Neutral

(d) Unimportant

(e) Very Unimportant

How impartant is the proposition "The University must be given the
right to expel any student convicted of a crime committed during a
violent demonstration"?

 

(a) Very important L
(b) Important

(c) Neutral ,

(d) Unimportant 3

(e) Very Unimportant 5

How impartant is the proposition "The University must take a strong \

institutional stand against American participation in the Viet Nam
war"?

(a) Very important
(b) Important

(c) Neutral

(d) Unimportant

(e) Very unimportant

How impartant is the proposition "University police must be prohibited
from carrying firearms on campus"?

(a) Very important
(b) Important

0 Neutral

(d) Unimportant

(e) Very unimportant

How impartant is the proposition "The University should prohibit
all military recruiting on campus"?

(a) Very important
(b) Important

(c) Neutral

(d) Unimportant

(e) Very unimportant

APPENDIX B

EXPERIMENTAL QUESTIONNAIRE

 

 

75
OPINION PROFILE
INSTRUCTIONS

The purpose of this profile is to obtain your opinions regarding the
probable truth and the desirability of a proposition. In the following
pages you will find a proposition and several messages, followed by a
series of questions. For example, here is a proposition similar to the
one you will see:

SAMPLE PROPOSITION
Detergents are polluting our nation's rivers.

MESSAGES

Two messages will then comment on the proposition. The first message
will support the proposition; the second message will attack the proposi-
tion.

QUESTIONS

The first set of questions will ask you to write down the important
aspects of both the attack and support messages.

The next question will ask you to rate the probable truth of the
proposition on a scale ranging from 0% to 100% true. Please place an "X"
in the blank that most accurately describes your evaluation of the truth
of the proposition.

The following question will ask you to rate the desirability of the
proposition on a scale ranging from."Very desirable" to "Very undesirable".
Once again, please place an "X" in the blank that most accurately
describes your estimate of the desirability of the proposition.

The last question will ask you to write a one paragraph statement
indicating why you thought the proposition was or was not true, and why
you thought the proposition was or was not desirable.

ORDER

Please complete all of one page before proceeding to the next page.
Once you have completed a page please do not return to it. You may,
however, return to this page at any time in order to clarify any instruc-
tion that is unclear to you.

“KS-

 

76

Propasition - University administrators must be given the right to search
student dormitory rooms for drugs.

The reason that some people believe this is that narcotics are drugs.
According to Dr. Joseph Elder, professor of sociology at the University of
Wisconsin, narcotics are legally defined as addictive drugs. Addictive
drugs may appear in many forms--pills, liquids, powders, etc. According to
Peter Leonard, associate director of the National Education Association
(N.E.A.), student dormitory rooms are temprorary domiciles in which stu- E?
dents keep books, clothes and other personal belongings. Thus, university

administrators must be given the right to search student dormitory rooms.

 

Some people, however, believe that university administrators must not “.
be given the right to search student dormitory rooms because administra-
tors are systems analysts. According to Dr. Joseph WOelfel, professor of
sociology at the University of Illinois, university administrators are
normally defined as the president, provost, registrar and deans of the
various colleges. James Small, research assistant for the American
Association of University Professors (A.A.U.P.), argues that an educator
functioning as a systems analyst must have expertise in the social,
psychological and maintenance systems of universities. Thus, university
administrators must not be given the right to search student dormitory

rooms 0

77

Propgsition - University administrators must be given the right to search
student dormitory rooms for drugs.

The reason some people believe this is that narcotics are drugs.
According to Dr. Joseph Elder, professor of sociology at the University
of Wisconsin, narcotics are legally defined as addictive drugs. Addictive
drugs may appear in many forms--pills, liquids, powders, etc. According
to Peter Leonard, associate director of the National Education Association
(N.E.A.), student dormitory rooms are temporary domiciles in which stu-
dents keep books, clothes and other personal belongings. Thus, university
administrators must be given the right to search student dormitory rooms.

Some people, however, believe that university administrators must not
be given the right to search student dormitory rooms because this would be
a violation of individual rights. According to Dr. Joseph Weelfel, pro-
fessor of sociology at the University of Illinois, the right to search a
student's dormitory room would violate the student's right of freedom.from
unlawful search and seizure. James Small, research assistant for the
American Association of University Professors (A.A.U.P.), argues that
student might come to fear, rather than respect, the authority of the
administration if it were allowed to search student dormitory rooms.

Thus, university administrators must not be given the right to search

student dormitory rooms.

..H "

 

 

78

Propgsition - University administrators must be given the right to search
student dormitory rooms for drugs.

The reason that some people believe this is that narcotics are drugs.
According to Dr. Joseph Elder, professor of sociology at the University of

Wisconsin, narcotics are legally defined as addictive drugs. Addictive

 

drugs may appear in many forms--pills, liquids, powders, etc. According

to Peter Leonard, associate director of the National Education.Association
(N.E.A.), student dormitory rooms are temporary domiciles in which stu- I3-
dents keep books, clothes and other personal belongings. Thus, university' I

administrators must be given the right to search student dormitory rooms.

 

However, some people believe that university administrators must not ~11
be given the right to search student dormitory rooms because most students 5
do not store drugs in their rooms. According to Dr. Joseph Woelfel, pro-
fessor of sociology at the University of Illinois, a study of three
Illinois colleges revealed that none of the students interviewed had
used their rooms to store drugs. James Small, research assistant for the
American Association of University Professors (A.A.U.P.), found in a
nationwide study that none of the 800 students he interviewed had stored
drugs in their dormitory rooms. Thus, university administrators must

not be given the right to search student rooms.

79

Proposition - University administrators must be given the right to search
student dormitory rooms for drugs.

The reason some people believe this proposition is that one conse-
quence of the early identification and counseling of drug users is that
they are less likely to become serious drug abusers. According to Dr.
Joseph Elder, professor of sociology at the University of Wisconsin, the
initiation of an effective drug user identification program would lead to
a sharp decline in the number of student drug addicts. Peter Leonard,
associate director of the National Education Association (N.E.A.), states
that if students can be identified in the early stages of drug usage,
educational and psychological counseling can frequently prevent drug
addiction. Thus, university administrators must be given the right to
search student dormitory rooms.

Some people, however, believe that university administrators must not
be given the right to search student dormitory rooms because administrators
are systems analysts. According to Dr. Joseph Woelfel, professor of
sociology at the University of Illinois, university administrators are
normally defined as the president, provost, registrar and deans of the
various colleges. James Small, research assistant for the American
Association of University Professors (A.A.U.P.), argues that an educator
functioning as a systems analyst must have expertise in the social, psy-
chological and maintenance systems of universities. Thus, university
administrators:must not be given the right to search student dormitory

rooms a

 

8O

Propasition - University administrators must be given the right to search
student dormitory rooms for drugs.

The reason some people believe this proposition is that one conse-
quence of the early identification and counseling of drug users is that
they are less likely to become serious drug abusers. According to Dr.
Joseph Elder, professor of sociology at the University of Wisconsin, the
initiation of an effective drug user identification program would lead to
a sharp decline in the number of student drug addicts. Peter Leonard, $2
associate director of the National Education Association (N.E.A.), states 1

that if students can be identified in the early stages of drug usage, t

 

educational and psychological counseling can frequently prevent drug \
addiction. Thus, university administrators must be given the right to
search student dormitory rooms.

Some people, however, believe that university administrators must
not be given the right to search student dormitory rooms because this
would be a violation of individual rights. According to Dr. Joseph
Wbelfel, professor of sociology at the University of Illinois, the right
to search a student's dormitory room would violate the student's consti-
tutional right of freedom.from.unlawful search and seizure. James Small,
research assistant for the American Association of University Professors
(A.A.U.P.), argues that students might come to fear, rather than respect,
the authority of the administration if it were allowed to search student
dormitory rooms. Thus, university administrators must not be given the

right to search student dormitory rooms.

81

Proposition - University administrators must be given the right to search
student dormitory rooms for drugs.

The reason some people believe this proposition is that one conse—
quence of the early identification and counseling of drug users is that
they are less likely to become serious drug abusers. According to Dr.
Joseph Elder, professor of sociology at the University of Wisconsin, the
initiation of an effective drug user identification program would lead to
a sharp decline in the number of student drug addicts. Peter Leonard, k
associate director of the National Education Association (N.E.A.), states

that if students can be identified in the early stages of drug usage,

 

 

educational and psychological counseling can frequently prevent drug \‘
addiction. Thus, university administrators must be given the right to
search student dormitory rooms.

However, some people believe that university administrators must not
be given the right to search student dormitory rooms because most students
do not store drugs in their rooms. According to Dr. Joseph WOelfel, pro-
fessor of sociology at the University of Illinois, a study of three
Illinois colleges revealed that none of the students interviewed had used
their rooms to store drugs. James Small, research assistant for the
American Association of University Professors (A.A.U.P.), found in a
nationwide study that none of the 800 students he interviewed had stored
drugs in their dormitory rooms. Thus, university administrators must not

be given the right to search student dormitory rooms.

III 1 all. I

82

Propoaitioa - University administrators must be given the right to search
student dormitory rooms for drugs.

The reason some people believe this proposition is that students
have been found to store drugs in their dormitory rooms. According to
Dr. Joseph Elder, professor of sociology at the University of Wisconsin,
inpdepth interviews with twenty-two students who were treated for drug
addiction revealed that student rooms were among the most frequently used
places for storing drugs. Peter Ieonard, associate director of the #5
National Education Association (N.E.A.), conducted a nationwide survey

which produced similar results. His survey of 1,200 college drug users

 

revealed that 68.5% of the students listed student rooms as one of the \
most commonly used places for storing drugs. Thus, university adminis-
trators must be given the right to search student dormitory rooms.

Some people, however, believe that university administrators must not
be given the right to search student dormitory rooms because administra-
tors are systems analysts. According to Dr. Joseph Whelfel, professor of
sociology at the University of Illinois, university administrators are
normally defined as the president, provost, registrar and deans of the
various colleges. James Small, research assistant for the American
Association of University Professors (A.A.U.P.), argues that an educator
functioning as a systems analyst must have expertise in the social, psyb
chological and maintenance systems of universities. Thus, university
administrators must not be given the right to search student dormitory

rooms 0

83

Proposition - University administrators must be given the right to search
student dormitory rooms for drugs.

The reason some people believe this proposition is that many stu-
dents have been found to store drugs in their dormitory rooms. According

to Dr. Joseph Elder, professor of sociology at the University of Wisconsin:
in—depth interviews with twenty-two students who were treated for drug

addiction revealed that student rooms were among the most frequently used
places for storing drugs. Peter Leonard, associate director of the
National Education Association (N.E.A.), conducted a nationwide survey
which produced similar results. His survey of 1,200 college drug users
revealed that 68.5% of the students listed student rooms as one of the
most commonly used places for storing drugs. Thus, university administra-
tors must be given the right to search student dormitory rooms.

Some people, however, believe that university administrators must not
be given the right to search student dormitory rooms because this would be
a violation of individual rights. According to Dr. Joseph Woelfel, pro-
fessor of sociology at the University of Illinois, the right to search a
student's dormitory room would violate the student's constitutional right
of freedom from.unlawful search and seizure. James Small, research assis-
tant for the American Association of University Professors (A.A.U.P.),
argues that students might come to fear, rather than respect, the authority
of the administration if it were allowed to search student dormitory rooms.
Thus, university administrators must not be given the right to search

student dormitory rooms.

 

84

Proposition - University administrators must be given the right to search
student dormitory rooms for drugs.

The reason some people believe this proposition is that many stu-
dents have been found to store drugs in their dormitory rooms. According
to Dr. Joseph Elder, professor of sociology at the University of Wisconsin,
in-depth interviews with twenty-two students who were treated for drug
addiction revealed that student rooms were among the most frequently used
places for storing drugs. Peter Leonard, associate director of the
National Education Association (N.E.A.), conducted a nationwide survey

which produced similar results. His survey of 1,200 college drug users

 

revealed that 68.5% of the students listed student rooms as one of the Rt
most commonly used places for storing drugs. Thus, university adminis-
trators must be given the right to search student dormitory rooms.
However, some people believe that university administrators must not
be given the right to search student rooms because most students do not~
store drugs in their rooms. According to Dr. Joseph Whelfel, professor
of sociology at the University of Illinois, a study of three Illinois
colleges revealed that none of the students interviewed had used their
rooms to store drugs. James Small, research assistant for the American
Association of University Professors (A.A.U.P.), found in a nationwide
study that none of the 800 students he interviewed had stored drugs in
their dormitory rooms. Thus, university administrators must not be given

the right to search student dormitory rooms.

85

In supporting the above proposition, Dr. Elder said

 

 

 

Peter Leonard said

 

 

 

In attacking the above proposition, Dr. Woelfel said

 

 

 

James Small said

 

 

 

 

1'" _‘"

1.

86

How true is the proposition that: "University administrators must
be given the right to search student dormitory rooms for drugs"?

(a) 0 - 10% true
(b) 11 - 20% true
(C)_____21 - 30% true
(d) 31 - 40% true
(e)_____41 - 50% true
(f) 51 - 60% true
(g)__61 - 70% true
(h) 71 - 80% true

(i) 81 - 90% true
(J)____91 400% true

How desirable is it that university administrators be given the right
to search student dormitory rooms for drugs?

(a) Very desirable
(b) Desirable

(c) Undesirable

(d) Very undesirable

Compose a one paragraph essay which states the reasons for your
answers to questions Q and tag.

 

 

87

DEMOGRAPHIC INFORMATION

What is your age? years

What is your religion?

 

 

 

( ) Protestant ( ) Catholic ( ) Jewish ( ) Other
(specify)

Are you... ( ) Male ( ) Female
What is your college class?
( ) Freshman ( ) Sophomore ( ) Junior ( ) Senior
What is your cumulative grade point average (GPA)?
( ) 3051 - 4.00
( ) 2.51 - 3.00
( ) 2.00 or less
What is your major?
What are the education levels of your mother and father?

Father Mother
Grade school or less (grade 1 - 8) 1 1
Some high school 2 ( ) 2 ( )
Graduated high school 3 ( ) 3 ( )
Some college 4 ( ) 4 ( )
Graduated college 5 ( ) 5 ( )
Post-graduate work ' 6 ( ) 6 ( )

Have you ever taken a course in persuasion or logic?

( ) Yes ( ) No