An experimental study of campaign issues and candidates' personality traits as influencing variables on voting behavior.

MISSING IMAGE

Material Information

Title:
An experimental study of campaign issues and candidates' personality traits as influencing variables on voting behavior.
Physical Description:
v, 63 leaves. : ill. ; 28 cm.
Language:
English
Creator:
Kitchens, James Travis, 1949-
Publication Date:

Subjects

Subjects / Keywords:
Voting   ( lcsh )
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis--University of Florida.
Bibliography:
Includes bibliographical references.
Statement of Responsibility:
By James Travis Kitchens.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 000580885
notis - ADA8990
oclc - 14094141
System ID:
AA00003539:00001

Full Text













AN EXPERIMENTAL STUDY OF CAMPAIGN ISSUES
AND CANDIDATES' PERSONALITY TRAITS AS
INFLUENCING VARIABLES ON VOTING BEHAVIOR












By


JAMES TRAVIS KITCHENS


A DIPRFr:rAT*. o PI,':%~) 7I
THE UNIVERSITY)
IN PARTIAL FULFILLMtI.fT OF
DEGREE 0- DOCriR


) THE GRADUATE COUNCIL OF
SOF FLORIDA
THE REQUIREMENTS FOR THE
OF PHILOSOPHY


1974















ACKNOWLEDGEMENTS

The author wishes to express his thanks to Dr. Thonas J. Saine,

Dr. Donald E. Williams, Dr. Anthony J. Clark, and Dr. Alan Agresti for

their help during this study. The author also wishes to express

special thanks to Dr. Douglas G. Bock, chairman, for his constant

guidance and advice without which this study would not have been

possible.

To his wife, Linda, and son, Jamie, the author wishes to express

his thanks for their love, patience, and tolerance during the completion

of this work.

Finally, thanks to HAL for not ABENDING.
















TABLE OF CONTENTS


Page

ACKNOWLEDGEMENTS .. ... . .......... ii

LIST OF TABLES. . ... . iv

ABSTRACT .. .. . ............ vI

CHAPTER I: STATEMENT OF THE PROBLEM AND REVIEW OF THE LITERATURE 1

CHAPTER 11: METHODOLOGY. . .... .. 18

CHAPTER III: RESULTS. . ... 29

CHAPTER IV: CONCLUSION AND DISCUSSION. . 41

APPENDIX A: DESCRIPTION OF THE CANDIDATES. . ... 55

APPENDIX B: TV ADS. . . ... ..... 57

APPENDIX C: RADIO ADS. . ... ..... 59

APPENDIX D: THE CITIZEN'S BUREAU REPORT ON THREE CONGRESSIONAL
CAMPAIGNS. . . ... 60

APPENDIX E: THE CONGRESSIONAL PREVIEW. . ... .62

BIOGRAPHICAL SKETCH. . . ... .. 63















LIST OF TABLES


Table Page

1. Issues and Trait Valence from the Offset Data 30

2. Prediction Functions from the Offset Data and Variables 31
Used in the Functions with DI Indicating Agreement or
Desirable Dimension and D2 Indicating Importance
Dimension

3. Comparison of Predicted Percentages Using Pretest Pre- 32
dictions and Actual Vote Count from the Experimental
Groups

4. Misclassification of Predictions Using Pretest Data 32

5. Association of Variable Rank for Offset Data and Exper- 33
mental Posttest Data Using Spearman's r

6. Comparison of Issue, Personality, and Combination 33
Discriminant Models Using Posttest Data

7. Classification from the Discriminant Analysis Pretest 35
and Posttest (Ambivalent)

8. Important Variables from the Pretest Discriminant 35
Analysis and Means for Candidate Voter Groups
(Ambivalent)

9. Important Variables from the Posttest Discriminant 35
Analysis and Pretest and Posttest Means for Candidate
Voter Groups (Ambivalent)

10. Classification from the Discriminant Analysis Pretest 36
and Posttest (Positive)

11. Important Variables from the Pretest Discriminant 36
Analysis and Means for Candidate Voter Groups
(Positive)

12. Important Variables from the Posttest Discriminant 36
Analysis and Pretest and Posttest Means for Candidate
Voter Groups (Positive)









Table


13. Within Groups Variance for Aggressive and Impeachment 37
Pretest and Posttest

14. Classification from the Discriminant Analysis Pretest 38
and Posttest (Negative)

15. Important Variables from the Pretest Discriminant 38
Analysis and Means for Candidate Voter Groups
(Negative)

16. Important Variables from the Posttest Discriminant 38
Analysis and Pretest and Posttest Means for Candidate
Voter Groups (Negative)

17. Control Group t's for Issue and Personality Variables 39

18. Means for Agreement and Desirability Dimensions for the 47
Issue-Only Model and Personality Traits-Only Model in
the Ambivalent Case with Candidate A Being Positive
on Issues and Candidate P Being Positive on Personality
Traits


Page















CHAPTER 1


STATEMENT OF THE PROBLEM AND
REVIEW OF THE LITERATURE

Attempts to understand the functioning of political communication

have attacked the problem from many different approaches, including

political science concepts, sociological concepts, and communication

concepts. The main problem with research in this area is that the dis-

jointed nature of the approaches has not led to a systematic theory

building and theory testing process. Studies of political communica-

tion and voting behavior have located many variables that are related

to the voting process. Kitt and Gleicher investigated the influence

of peer relationships on voting according to party preference. They

concluded that "if a respondent is a member of mixed friendship groups

[both Democrats and Republicans] he is less likely to feel
2
strongly about his vote inclination." Shepard constructed a model of

voting behavior comparing economic self interest and public regarding-

ness. He stated that "a model of referendum voting as a behavior based

upon calculations of economic self interest was formulated and found to

yield empirically congruent predictions with four times the frequency

of predictions developed from a model of public regardingness as a

value premise in voting behavior."

Two of the important variables which have emerged are issues and

candidates' personalities. For example, Eli found that the three

variables of candidate's image, party image, and issue orientation









accounted for .98 of the variance in his model of voting behavior in

the 1970 Tennessee gubernatorial election.5

Issues are not defined in the sense that the term is used by some

political candidates. That is, many times any variable which might

affect the voters' decision is labeled "issue." In terms of political

communication, and in this study, an issue is defined as a statement of

proposed governmental policy, a defense of present governmental policy,

or an attack of present or proposed governmental policy.

Personality traits are defined as consistent behavior by the candi-

dates which indicate mental or emotional traits. Personalities are

indicated by any available information about the candidates' personal-

ity traits. The reality of these traits is unimportant, except as they

exist in the voters' minds.

The question, which is and has been debated, concerns the relative

importance of these two variables in the voters' decisions. Gene

Wyckoff argued that the image or personality of the candidate is the

most important factor: "The influence of issues on the outcome of

elections also seems to be declining as 1) political questions become

too complex for ready statement or comprehension, and 2) candidates

themselves avoid assuming issue positions that might be considered too

extreme."

Contrary to this view, Sherrod observed that the consideration of

governmental policy is the important variable in the voters' judgments:

"the less intense their voters' feelings toward the issue, the greater

is their tolerance for inconsistency between their own and their candi-

date's position," and vice versa.

In his study of the 1972 Presidential campaign, Swanson found that










voters listed solutions to problems and personal qualities as the two

most important considerations in their decision. The dominance of one

factor, however, was unclear. Of the Nixon voters 27.4% said that per-

sonal qualities were their main consideration and 34.5% of the Nixon

voters reported that his solutions to problems were the primary basis

for their decision. Of the McGovern voters who were surveyed, 59.0%

said their decision was based on personal qualities of the candidate

and 27.2% based their decision on McGovern's approach to problems.

Before attempting to understand how these two variables work,

there is a basic question that needs to be answered. Are these two

variables evaluated similarly by the voter? That is, does the voter

use the same frame of reference to evaluate the two variables? Osgood,

Suci, and Tannenbaum investigated the existence of a political frame of

reference. They reported:

This indicates, according to our interpretation, that despite
their different political outlooks, despite their gross dif-
ferences in the meanings of particular concepts, like TRUMAN,
OUR POLICY IN CHINA, and SENATOR McCARTHY, these groups of
voters employ essentially the same frame of reference in
making political judgments. They have the same sets of
"values," the same relevant discrimination with respect to
political persons and issues.9

The discussion of the structure of the political frame of reference will

be presented later in a discussion of the nature of political attitudes.

The important point in Osgood, Suci, and Tannenbaum's research is that

there is a consistent frame of reference for both issues and political

personalities.

Perhaps the most intensive study of any political communication was

the study of the Nixon-Kennedy television debates in 1960.10 The series

of studies attempted to locate some important variables which affected

the electorates' perceptions of the candidates. Although many









variables such as channel of communication, candidates' physical appear-

ance, and issue positions, caused speculation, a general model of the

judgmental process was not presented. Without a general model of this

process, an orderly investigation of important variables, the inter-

action of variables, and new concepts that are discovered may not be

made, since the direction needed to understand political communication

will be unclear.

Two dissertations written at Southern Illinois University attempted

to use linear regression in two gubernatorial campaigns.1 The models

had three variables; issues, candidates' personalities, and social

factors. Both models were created after the election. Therefore,

although the models gave an indication of the weight of certain vari-

ables, little evaluation of the effects of the communication process

could be made.

Most studies of political communication and voting behavior have a

common approach. The information has been gathered in field settings and

related to actual elections. This empirical method has made valuable

contributions in the location of important variables. The main problem,

however, is that such studies lack any control which is necessary for

an in-depth and isolated examination of variables. Through a system-

atic controlling of political communication variables, the manner in

which a variable operates can become clear because outside sources of

variation can be controlled or eliminated. With an understanding of how

a limited number of variables operate, new political communication

variables can be added to the controlled situation to determine inter-

actions within the process, thus building a model of the process. The

purposes of this study were to determine if issues or personality






5


traits were the dominant factor in the voter's decision, and if the re-

inforcement of existing attitudes through communication caused attitude

change as shown by the use of a controlled election situation.

There were two questions which were investigated in this study.

First, was there a dominance of one of the two variables in the voters'

decisions? To explore this question, the functioning of these two

variables were examined when the voters were choosing between two liked

candidates, two disliked candidates, and two ambivalent candidates.

Secondly, what was the effect of the campaign process on the

voters' attitudes? To answer this question, a comparison of the impor"

tant variables of voters' decisions in an immediate choice situation was

compared to the important variables in a choice made after the exposure

to the campaign communication process. Also, an examination of atti-

tudes before and after the campaign was done to determine if the re-

inforcement of attitudes changed the degree of the attitude. To further

explain and justify these research questions, it is necessary to examine

the nature of political attitudes, a theory of the political campaign as

a persuasive process, and the use of discriminant analysis.

The Political Attitude

Research on the measurement of attitudes has suggested that atti-
12
tudes are not unidimensional.12 Osgood, Suci, and Tannenbaum explored

the nature of political preference and suggested a two dimer 31 view
13
was necessary. The dimension, other than agree-disagree, seems to be

importance of a concept. In their discussion of attitudes as predictors

of behavior, Kiesler, Collins, and Miller observed:

Indeed, Rokeach and Rothman found that predictions are signi-
ficantly better when one takes account of the importance of
separate items. The more important the subject thinks X is,









say, the more it affects his evaluation of X and Y in combin-
ation. Implicatively, the more important the source is to the
subject, the greater the change in attitudes about the object
and the less about the source. The reverse would presumably
be true, the more important the object.l

Himmelstrand also explored the structure of attitudes as predictors. In

his study he found that "the disposition to make a transition for 'word,'

i.e., verbal attitudes, to action varies from individual to individual

depending on how deeply 'rooted' or 'anchored' verbal attitudes are in

the individuals personality or in other attitude components."15

There have been several studies of political communication using

the multi-dimensional view of political attitudes. Richard S. Elster

and James R. Capra created a two dimensional inodel by using similarity-

dissimilarity judgments of 88 subjects. They reported, "the results

suggest it might be possible to sort out the dimensions of appeal of
16
political personalities."6 Kitchens used a two dimensional model to

examine ego-involvement and importance of campaign issues in the 1972

Presidential campaign. The study suggested that there may be a corre-

lation between the two dimensions. The results indicated that "the

groups generally showed high ego-involvement with the highly ranked

questions based on the importance dimension."17

In Russell's investigation of media and non-media influences on

voting behavior in the 1970 gubernatorial campaign, many different

dimensions of attitudes were investigated. Russell explained that

"personal data about each subject was gathered to aid in determining

what, if any, cross pressures may combine with the media variables and

images to produce change in the subject."18

Eli also concluded multi-dimension measurement was needed to exam-

ine political attitudes. He stated, "it is the contention of this










researcher that there are probably several dimensions of any given atti-

tude toward an object or situation."19 In his study of the 1970

Tennessee gubernatorial election, Eli used semantic differentials,

behavioral differentials, and Likert-type scales.

There is a unique characteristic of political attitudes which

should be pointed out. The main goal of most attitude theories is the

connection between attitudes and behavior. Most criticism of attitude

theories is based on the fact that individuals do not behave toward

objects in the predicted direction of the theory, or that the theory is

not totally generalizable.20 The measurement of attitudes concerning

political factors does not try to predict the way a person will act when

presented with an object. That is, regardless of a subject's attitude

about the Arab-Israeli dispute, the person, himself, will never be

required to actually face the problem. Rather, a person will select

someone to face the problem for him, that person being the political

candidate. Therefore, whether or not a person would carry out his atti-

tude or not is unimportant. What is important is the belief that the

person that he is voting for has the same attitude and will carry out

the action. Further discussion of this point will be presented in an

explanation of the communication process of a political campaign.

The Campaign As A Persuasive Process

Many retrospective political communication studies of election

have been done. This approach yields few generalizations about the

effect of the campaign process. These studies, however, did not

attempt to control communication variables for two reasons. First, if

control were exerted, the researcher would become an intervening vari-

able interacting with the phenomena being studied. Secondly, the









pragmatics of a political campaign would restrict the level and amount

of control given any researcher. Although it would be difficult to

duplicate a political campaign in an experimental setting, a simulation

without losing necessary controls is possible.

The concept of using a simulation for the study of social phenomena

has been previously investigated. Dawson explained the desirability of

using a simulation methodology by stating:

In many instances, especially in the study of social and psy-
chological phenomena, it is undesirable or even impossible to
conduct experiments on real systems. By successfully simulat-
ing the significant variables, it is possible to explr e such
phenomena by experimenting with the simulated system.

Although there have been several attempts to apply'a simulation
22
methodology to the voting situation,2 Laponce pointed out that this

methodology has been somewhat overlooked:

Though many political scientist, as advisers to governments,
have been in a favourable position to perform realistic sim-
ulations, they have either refrained from doing so, because
they were not able to convince the politicians with whom
they had to collaborate, or more likely, had not thought of
it. 3

Deutsch took a more extreme position that the utilization of the

simulated methodology is not only more desirable, but necessary to

understand the political situation:

The truth is not in any one kind of evidence but in the
relation of many kinds. The research for multiple evidence
may require laboratory experiments, survey research, content
analysis, general observations and case studies, individual
and aggregative data, and therefore five or six kinds of
evidence fpr every serious effort to understand what is
going on.

The necessity of including a simulation of the communication

process can also be supported by the fact that voters' decisions occur

during different times in the campaign. For example, survey data on

Presidential elections from 1948 to 1968 indicated that voters










reached their decisions at different times during campaigns.25 Swanson
26
further confirmed this notion in his study of the 1972 campaign.2

These empirical reports add support to Fotheringham's theory of persua-

sion: "The concept of instrumentality encourages seeing persuasion as a

campaign a structured sequences of efforts to achieve adoption, con-

tinuance, deterrence, or discontinuance rather than as a one-shot

effort."27

There seem to be two possible reasons for voters' decisions being

formulated at different times. First, the difference may be due to

inherent individual differences in each person's cognitive ability to

make decisions. Secondly, it may be a socialization factor. Although

there is a deadline for the decision to be made, when the voter enters

the voting booth, the voter knows that he may make the decision, and

evan change it several times, at any point during the time period.

Either way, by allowing a time period for decision with subjects aware of

a deadline, a simulation of the decision-making process as well as a

simulation of the campaign structure could be constructed. The specific

controls of the situation will be discussed in Chapter Two.

The concept of persuasion in political communication must be examined

from two viewpoints. First an examination of a theory of persuasion

from the view of the communicator is necessary. Next, a view of attitude

change and processing of the communication will be examined from the

receiver's point of view. An examination of these two theories can

reveal how they are related in terms of political communication, and in

fact, may be one theory.

Persuasion in the modern political campaign does not aim for a

change in the type of response that the voter exhibits toward a concept,









but tries to reinforce that response. The candidate will attempt to

find, for example,what issue is important to the voters and attempt to

reinforce that belief in order to become acceptable to the electorate.28

Miller claimed that the reinforcement of attitudes is a goal of persua-

sion. He explained that "their evaluative responses need not be altered

in kind, only in degree the communicative end is the extension of

rewarding stimuli to the audience, an end calculated to increase the

probability that they will retain their present store of evaluated

responses."29

The reinforcement of existing attitudes does not mean that the aim

of political communication is stagnation. As Swanson explained, "Pre-

existing perceptions of candidates and their solutions to problems thus

do not preclude the possibility of change induced through political

communication; they merely set parameters within which the political

persuaders must work."30 This view of political communication leads to

the question of what happens in terms of the receiver.

Although the issue stand or personality trait that is presented to

the voter may not reach an individual's ideal of the concept, the posi-

tion may be acceptable to the individual. In their book Social Judg-

ment, Sherif and Hovland discussed this concept in terms of latitudes

of acceptance and latitudes of rejection.31 They defined the concept

stating:

A latitude of acceptance for certain stands on an issue im-
plies a rather definite range of rejection as well. It is
defined operationally as the range of the positions on an
issue that an individual considers acceptable to him (includ-
ing the one "most acceptable" to him). The latitude of re-
jection consists of the positions he finds objectionable
(including the one "most objectionable" to him).32

The two regions are divided by a non-committal region for those solutions

which have not yet been judged.









If this theory is examined in terms of political communication with

the persuasion theory, their relationship becomes clear. Sherif and

Hovland noted the differences in individuals with respect to their

theory; "The stands of different individuals vary in their personal

significance and in the size and locations of the latitudes of accept-

ance and rejection."33 Given the individual differences, the political

communicator must find an attitude position which is in the latitude of

acceptance of, theoretically, 50% plus one person in the electorate.

His communication must reinforce that position because of the unique

connection between political attitudes and the desired behavior, which

was previously discussed.

For example, the logic from the viewpoint of the candidate would

be, "I believe in A, A is acceptable to you, I like B, B is acceptable

to you, and I want to do C, C is acceptable to you, therefore VOTE FOR

ME!" The communication would be constructed in the negative when

speaking of an opponent.

Movement or change in attitudes From reinforcement could take one

of several forms. First, the continuing reinforcement of an acceptable

idea may make that solution "most acceptable" and hence, increase the

importance of the variable. Likewise, when a point in the latitude of

rejection is attacked, the unacceptable idea may become the "most

unacceptable."

Conversely, if points in the latitude of acceptance are attacked,

the receiver will become stronger in his own belief, and will align

himself further away from the advocated stance. Also, if the communi-

cator attempts to reinforce in a positive manner points in the latitude

of rejection, the receiver will see those points as more negative.








Sherif and Hovland called the functioning of variables in this manner

the assimilation-contrast effect.

Finally, when a dichotomous choice has to be made, it is possible

that the uncommitted will accept the concept. Sherif and Hovland sup-

port this concept; "When an individual adopts a stand on a controversial

issue, he is mindful of the prevailing opinions in the group of which he

is a member or which he aspires."35

To summarize, political communication is a persuasive process

which functions by reinforcing known existing acceptable attitudes.

Although the variables may be altered during the process, it may be

possible to use a sample to locate acceptable concepts and predict

voting behavior based on the set of concepts without knowing the exact

order of significance. If there is a pattern of change which can be

discovered, it will indicate two important concepts. First, if there is

a pattern, it will indicate a consistent functioning of these two vari-

ables in voting behavior. This information would be the first step in

constructing a model of voting behavior. Secondly, with equal reinforce-

ment of all the variables, if a pattern emerges, then to make political

communication more efficient, the amount of reinforcement on these

variables could be increased.

Discriminant Analysis

To determine which variables were most important and if there was

a pattern in the way voters decide on issue and personality traits,

a discriminant analysis was performed on the attitude data.36 Discrim-

inant analysis is a multivariate technique which uses a density

function in order to place subjects into different groups. To estab-

lish a model for a group, it is necessary to know which group the data

initially came from.










For example, to establish a model to discriminate between candidate

X's voters and candidate Y's voters, the attitude data from a group of

subjects who voted for each of the candidates would be placed under

their respective groups. A discriminant analysis would then be applied.

This analysis would produce several important facts. First, the results

will reveal how many misclassifications of subjects result from the use

of the density function. Secondly, the variables are ranked in order of

importance in the discrimination, and an F score is given for each vari-

able to determine if the variable is independently significant.

Finally, a discriminant linear function in stepwise fashion for each

group is given which can be used to classify new data into each group.

Hence, this function can be a predictor.

New data can be classified by applying the new data to the func-

tion:

assign to group 1 if z'C]+C10o z'C2+C20, otherwise assign to group 2,
where
z' = data matrix
Cx = constant vector
Cx0 = constant

The probability that a new piece of data has been misclassified can be

calculated with the formula:


1 ez'Cl+C10
2
Z- eZ'Cl+C10
G=l


The error of misclassification gives more information about the good-

ness of the fit of the model.

Research Hypotheses

Working from the concepts of social judgment theory, the following

research hypothesis were formulated:









1) Because pre-existing attitudes were reinforced as opposed to

attempting to alter the kind of attitudinal responses, the variables

which were important in discriminating between the voting groups in the

offset data will be the same variables important for discriminating

between voter groups in the posttest data.

2) Since the variables important for the discrimination in the offset

data will be the same as the important variables in the posttest data,

the discriminant linear function produced by the discriminant analysis

computed for the offset data will produce accurate predictions of the

voting behavior for subjects in the experimental condition.

3) Because the importance of a variable is dependent upon the level of

ego-involvement as opposed to situational categories, such as issues and

personality traits, a combination of the variables will discriminate

between group of voters better than either variable alone.

4) Due to the assimilation-contrast effects resulting from the rein-

forcement of attitudes, voters attitudes, after experiencing the simu-

lation, will be more positive toward variables associated with the

selected candidate, and more negative toward variables associated with

the rejected candidate.

Implications and Limitations

By placing controls on the situation, an isolated look at two

important variables as they relate to the political decision can be

made. This type of examination cannot be made in a field study for two

reasons; 1) there is no way to control other variables, and 2) taking a

number of measurements for exploratory purposes is difficult because of

the size of the'sample needed for an accurate measurement of the popu-

lation.









Even with the control of variables there are certain limitations

which are inherent in the study. First, this study did not construct a

complete model of voting behavior. The two variables under examination,

however, do seem to be most important and the necessary starting point

for an understanding of the effects of political communication. Secondly,

the effects of different media was not explored in this study. The

process of political communication is a multimedia process, and the

question of the effects of each media is important to the study of

political communication.

Despite the limitations, this study can make two important contri-

butions. By the elimination of variables, except the two under examina-

tion, the functioning of these political communication variables as they

influence the voters' judgmental process can be made clear. From the

understanding of these variables, new variables can later be added, thus

continuing an orderly process of political communication model building.

Secondly, the simulation of the political communication process can

produce both a new methodology for future political communication theory

construction, and new knowledge about variables which can be examined

in actual field situations, thus lending continuity of direction to the

problem of understanding political communication.















Notes

1. See Hugh A. Bone and Austin Ranney, Politics and Voters,
(McGraw-Hill Book Company, New York, 1963), Sidney Kraus, ed., The
Great Debates (Indiana University Press, Bloomington, 1962),
Paul F. Lazarfeld, Bernard Berelson, and Hazel Gaudet, The People's
Choice (Columbia University Press, New York, 1944), and Angus Campbell,
Gerald Gurin, and Warren Miller, The Voter Decides (Row, Peterson, & Co.,
Evanston, 1954) for examples.
2. Alice S. Kitt and David B. Gleicher, "Determinants of Voting
Behavior," Public Opinion and Propaganda, (Dryder Press, New York, 1954),
p. 416-417.
3. William Bruce Shepard, Political Preference, Participation, and
Local Policy-Making: A Study of Referendum Voting Behavior in American
Cities (Ph.D. Dissertation, University of California, 1972).
4. For examples, see Gene Wyckoff, The Image Candidates: American
Politics in the Age of Television, (The Macmillan Company, New York,
1968), Joseph Napolitan, The Election Game and How To Win It, (Doubleday
and Co., New York, 1972), John Michael Cornett, A Descriptive Study of
Speaking in the 1968 Florida Democratic Senatorial Primary Campaign,
(Ph. D. Dissertation, Florida State University, 1969), Jack C. Eli, A
Study of Political Attitudes and Voting Behavior in the 1970 Guber-
natorial Election in Tennessee, (Ph. D. Dissertation, Southern Illinois
University, 1971), Charles G. Russell, A Multivariate Descriptive Field
Study of Media and Nonmedia Influences on Voting Behavior in the 1970
Texas Governor's Election, (Ph. D. Dissertation, Southern Illinois
University, 1971).
5. Eli, p. 90-91.
6. Wyckoff, p. 6.
7. D. R. Sherrod, "Selective Perception of Political Candidates,"
Public Opinion Quarterly, (Winter, 1971), p. 556.
8. David vi Swanson, "Political Information, Influence, and Judg-
ment in the 1972 Presidential Campaign," Quarterly Journal of Speech,
Vol. 59, (April),1973), p. 135.
9. Charles E. Osgood, George J. Suci, and Percy H. Tannenbaum,
Measurement of Meaning, (University of Chicago Press, Chicago, 1971),
p. 121.
10. See Kraus, The Great Debates.
11. See Elli, and Russell.
12. For examples, see Osgood, et. al., Stuart Kaplan, "The
Relationship Between Prominence and Valence of Perceived Properties of
Attitude Objects," Speech Monographs XXXVI, (1970), p. 278-281,
Carolyn S. Sherif, Muzafer Sherif, and Roger E. Nebergall, Attitude and
Attitude Change, (Saunders, Philadelphia, 1965), Donald J. Cegala and
Robert J. Kibler, "Object Importance and Commitment to Positions Pre-
dictors of Attitude Change," Central States Speech Journal XXIV,
(Summer, 1973), p. 108-116.









13. Osgood, et. al., p. 120-124.
14. Charles A. Kiesler, Barry E. Collins, and Norman Miller, Atti-
tude Change, (John Wiley and Sons, Inc., New York, 1969), p. 182-183.
15. Ulf Himmelstrand, "Verbal Attitudes and Behavior: A Paradigm
for the Study of Message Transmission and Transformation," Public
Opinion Quarterly, (Summer, 1960), p. 226.
16. Richard S. Elster and James R. Capra, "Multidimensional Scal-
ing of Political Preference," Perceptual and Motor Skills, (December,
1972), p. 990.
17. James T. Kitchens, "Presidential Campaign Issues: Ego-
Involvement and Importance," (Unpublished research paper, University of
Florida, 1972), p. 11.
18. Russell, p. 88.
19. Eli, p. 60.
20. Chester A. Insko, Theories of Attitude Change, (ppleton-
Century-Crofts, New York, 1967) and Kiesler, Collins, and Miller for a
criticism of attitude theories.
21. Richard A. Dawson, "Simulations in the Social Science,"
Simulations in the Social Sciences: Readings, Harold Guetkow, ed.,
(Prentice-Hall, Englewood Cliffs, N.J., 1962), p. 5-6.
22. For example see, Harold Guetkow, "Simulation in the Study of
Inter-Nation Relations," and Ithlel De Sola Pool and Robert Abelson,
"The Simulmatics Projects," in Simulations In the Social Sciences:
readings.
23. J. A. Laponce, "Experimenting: a Two-Person Game,"
J. A. Laponce and Paul Smoker, eds., Experimentation and Simulation In
Political Science, (University of Toronto Press, Toronto, 1972), p. 9.
24. Karl W. Deutsch, "The Contributions of Experiments Within the
Framework of Political Theory," Experimentation and Simulation in
Political Science, p. 31.
25. William H. Flanlgan, Political Behavior of the American
Electorate, (Allyn and Bacon, Boston, 1972).
26. Swanson, p. 133.
27. Wallace C. Fotheringham, Perspectives on Persuasion, (Allyn
and Bacon, Boston, 1966), p. 34.
28. Cornett, Chapter IV.
29. Gerald R. Miller, Speech Communication: A Behavioral Approach,
(The Bobbs-Merrill series in Speech Communication, New York, 1966),
p. 19-20.
30. Swanson, p. 133.
31. Muzafer Sherif and Carl I. Hovland, Social Judgment, (Yale
University Press, New Haven, 1961), Chapter 6.
32. Sherif and Hovland, p. 129.
33. Sherif and Hovland, p. 175.
34. For a review of assimulation-contrast research, see Kiesler,
Collins, and Miller, p. 240-297, and Chester A. Insko, Theories of
Attitude Change, (Appleton-Century-Crofts, New York, 1967), p. 64-91.
35. Sherif and Hovland, p. 125.
36. See Donald F. Morrison, Multivariate Statistical Methods,
(McGraw-Hill Book Company, New York, 1967) for a theoretical explana-
tion of discriminant analysis.















CHAPTER II


METHODOLOGY

The experimental design for this study was an offset pretest-

posttest design with a control group. The offset group was used as the

model group for the use of the discriminant analysis.

Mezasurei--nt of Attitudes

The response to issue statements was measured with a seven step

Likert--tpe scale ranging from "strongly agree" to "strongly disagree,"

and a seven step Likert-type importance scale, ranging from "very

important" to veryy unimportant."

The response to persn-ality traits was measured with a seven step

Likert-type scale ranging fro.n "highly dcLsirable" to "highly undesir-

abic," and a seven step Likrt-type importance scale, For numeric

analysis s, "strongly agree," "highly desirable," and "very important"

weie aesigned a value of 7, and "strongly disagree," "highly unjesir-

able," and "very unimportant" were assigned a value of 1.

The choice of issues was made by a survey of national rmlagazines,

national television newscasts, and newspapers. The issues used wrre

chosen based on the nu.iber of times that an issue was discussed in all

the media. Eight issues were chosen for the study. They included

1) the impeachment of the President, 2) the sale of arms to Israel,

3) the attempts of oil companies to repeal ecology measures, 4) the

limitation of Presidents to one term, 5) the return of price controls,

6) the Equal Rights Amendment, 7) ending the $750 child deduction on









Federal income tax, and 8) the public financing Presidential and Con-

gressional elections.

The choice of personality traits were selected from Osgood's

evaluative dimension and Hamilton and Huffan's study of impression for-

mation.l In order to put the personality traits in a political setting,

as opposed to simply stating the candidate has a certain trait, a pilot

study was conducted. Statements were written and the subjects (N = 60)

were asked to choose one of five traits which the statement indicated.

A blank space was also provided and the subjects were told that if they

felt none of the five adjectives described the statement adequately,

another adjective should be filled in. The statements were not accepted

unless at least 80% of the subjects agreed that the statement indicated

a certain trait.

From the study, eleven trait statements were found to be acceptable.

They were:

1) unaggressive- candidate X proposed much legislation while serving
as a state senator, but he did not push for their passage when they
mnt with opposition.

2) Inconsistent- candidate X changes his mind often. Sometimes on an
issue stand he has previously taken.

3) follyower- candidate X never proposed any original legislation, but
he supported many bills proposed by other senators.

4) dishonest- a citizens group investigated all the candidates in a
race. In investigating candidate X, they found evidence of cam-
paign funds coming from organized crime.

5) talker- candidate X has proposed many citizen committees to solve
local problems. However, he has never actively participated in the
organization of any of these groups.

6) progressive- candidate X helped raise funds to start an experimental
school that would use new techniques for the teaching of the
mentally retarded.

7) closeminded- candidate X has refused to support any legislation









which was not proposed by a member of the same political party as he
belongs to.

8) good natured- candidate X takes a lot of kidding about being bald,
but he responds by saying, "I'm not bald, other men are just too
hai ry."

9) aggressive- candidate X started his own business and made it into
one of the largest companies in the state. He has been elected to
the state legislature four times previously. He has never proposed
a bill in the legislature that has been defeated, although some of
them were quite controversial.

10) energetic- while in business, candidate X nmnaged three companies
at one time. He also found time to work as a "big brother" to
underprivileged children and head the United Way fund raising cam-
paign.

11) adventurous- candidate X spends his free time carping and traveling
by canoe down rivers all over the United States.

The Predictor Model

The attitude measurement of the issue statements and the personal-

Ity traits was given to a random sample of subjects (N = 33) in the

basic speech course at the University of Florida.2 The subjects were

asked to indicate tha degree of agreement with each issue statement and

to indicate how important, at that point in time, the issue would be to

them when considering whom to vote for in a congressional election. The

subjects were then asked to indicate how desirable each of the personal-

ity traits is for a political candidate, and how important it would be

for them to know about this trait before voting for a candidate for

Congress. This data will be labeled "model data" in the following

discussions.

A mean score for each issue and personality trait was computed. A

positive sign was placed on each item if the mean score was on the

"agree" or "desirable" end of the scale and a negative sign was placed

on an item if the mean score was on the "disagree" or "undesirable" end









of scale. Mean scores for the importance ratings were computed in order

to locate the order of importance for the issues and personality traits.

By using the positive and negative traits, a number of hypothetical

candidates were created. There were two candidates which were positive

on both issues and personalities, two candidates which were negative on

both factors, and two candidates which were ambivalent, with one candi-

date being positive on issues and negative on personality traits, and

the other candidate constructed oppositely.

For each candidate, two or three issues, and two or three person-

ality traits were used. The use of only a few ideas has become neces-

sary in modern American politics. Napolitan pointed out, "ordinarily a

candidate will have 10 to 15 ideas for programs. This is too many to

effectively utilize in a campaign. To use so many dilutes the message

of the candidate and causes some confusion in' the minds of the voters."3

The hypothetical candidates were then paired with two positive

candidates, two negative candidates, and two ambivalent candidates

competing. The subjects were given a written description of the pairs

of hypothetical candidates and told that the descriptions were of

candidates opposing each other in Congressional races in the near future.

The subjects were then asked to vote for one candidate in each race

(See Appendix A).

The subjects' attitude data on the variables used for each elec-

tion was grouped according to which candidate each subject voted for.

A discriminant analysis was then computed for the data to obtain

several pieces of information. First, the discriminant function that

was used as a predictor was obtained. Secondly, an estimation of the

goodness of the model was given. Finally, the important variables that









indicated the group distinctions based on the immediate choice situation

were located.

Prediction of Another Sample

Another random sample of subjects (N = 69) from the basic speech

course were given the attitude measurements of the same issue and

personality traits as the first group. This data will be referred to as

the pretestt data." The new data was placed into groups by applying the

discriminant function previously described to the data. The percentage

of subjects classified into each group was the predicted vote for each

candidate.

The Campaign As An Experimental Treatment

The subjects were told that they were going to select three men to

serve in the U. S. Congress in a mock election. The candidates, who

would actually be running in the near future, would spend five days

campaigning for their votes, and on the fifth day an election would be

held. The students were also told that the campaign and election would

be used later in the course in the study of communication and society.

Using the personality statements and the issue statements on the

attitude questionnaire, campaign messages were written. To insure that

no other concepts were introduced in the description of the personality

traits or issues, two validity checks were conducted using five judges.

To check for the validity of the personality trait statements, the

statements were presented followed by the same four adjectives pre-

sented in the original validity check of the personality traits plus a

space labeled "other." The judges were told to check all adjectives

which applied plus add any concept which was not on the list. No state-

ment was accepted unless four of the five judges agreed that only the

one desired concept applied.









To check the validity of the issues, the statements used were pre-

sented to five judges followed by four possible issues and a space

labeled "other." The judges were told to check the issues that they

felt were involved in each statement and list any other issue not on the

list. No statement was accepted unless four of the five judges agreed

that only the desired issue was involved.

During the five days, the campaign messages were presented to sub-

jects using video tape messages, taped messages to simulate radio, and

written messages. Each candidate had one video taped commercial, one

radio commercial, and two written messages (See Appendices B, C, D, and

E). One written message was labeled as being compiled by a citizens'

committee and the other report was labeled as an article by UPI repor-

ters. The written messages were not presented as campaign propaganda,

but contained information in report form. The video and audio messages

included only information about a candidate's positive points or his

opponents negative points.

On the fifth day of the campaign, the subjects voted using a simu-

lated voting booth. Each ballot was marked for identification with a

small number, and the subject's name who received the ballot was recor-

ded. This procedure was done in order to correlate the votes with the

attitude data. After each subject voted, he was given the attitude

questionnaire on the issues and personality traits again. This data

will be referred to as the "posttest data."

To explain the taking of the posttest, subjects were told that a

mistake had been made on the first questionnaire. They were told that

it was necessary for the person who created the questionnaire to know

whether or not each person had voted in an actual election. Since this

Information was not given on the original questionnaire, the forms









would have to be done again.


CAMPAIGN SCHEDULE*

1st day written message "The Citizens' Bureau Report On Three Con-

gressional Campaigns" (Appendix D)

2nd day television commercials (Appendix B)

3rd day radio commercials (Appendix C)

4th day written message "The Congressional Preview" (Appendix E)

5th day Voting and Posttest


Since the subjects were in their classes for an hour, each day
the ambivalent candidates' messages were presented at the
beginning of the period, the positive candidates' messages were
given near the middle of the period, and the negative candi-
dates' messages were given at the end of the period.


Experimertal Controls

This process approach i:iy seem to have lacked in control of the

independent variables when ccmpdred to many laboratory approach studies

in speech. If the reason for control and the types of controls which

were applied to the study are examined, however, then the process may

be seen as a legitimate approach. Kerlinger claimed that the control of

independent variables in a research design was "so that extraneous and

unwanted sources of systematic variance have minimal opportunity to

operate."

There were six controls which were placed on the process which

helped limit the probability of unwanted independent variables contami-

nating the study. The first three factors concern the candidates for

the mock elections. To control for acceptance or rejection of a candi-

date because of name or the association of a hypothetical candidate's

name with the name of an actual political figure, the hypothetical











candidates were assigned one letter labels (for example, candidate A,

candidate K, etc.). The second candidate variable controlled was contact

with the candidates. There were no pictures of any of the hypothetical

candidates, and the radio and video tapes were done in a third person

approach. Finally, there was no way for the subjects to gain any addi-

tional information about the candidates since they were hypothetical.

Another control which would be impossible in a field study was

the time of exposure to information. The time element for each candidate,

for each medium of presentation, and for each variable was held approxi-

mately equal. Also, for all the vocal messages, the same announcer

presented the messages, and on the video tape, he was not seen.

One danger in any study using realistic issues is that during the

study an event will happen that will influence or change the attitude

about one of the issues. To check for such an event, a control group

was examined. In a field study, a greater danger, perhaps, is the candi-

date doing something that changes the voter's view of the candidate's

personality. For example, Edmund Muskie's speech in New Hampshire and

Thomas Eagleton's background being exposed in the 1972 Presidential

campaign changed the voter's view of these men.5 Such an occurrence

was controlled in this laboratory situation.

With the concern over finding ways to control laboratory studies

in the social sciences, perhaps the danger of too much control has been

overlooked. Although control for the influence of unwanted variables

must be used, to obtain some realistic view of the process, some free-

dom of the variables under examination must be allowed. In this study,

unlike some communication studies, contact between the subjects would

not contaminate the study. Since this study was attempting to look at









a structure of a decision based on the exposure to a multimedia presen-

tation, including interpersonal communication, as opposed to attempt-

ing to explain the influence of a single media, contact between sub-

jects was necessary to the validity of the study.

Analysis of Results

For analysis of the research hypotheses, the subjects were grouped

according to the selected candidate in each race, and the following

statistical hypotheses were tested.

Hi: There is no significant difference in the predicted pro-
portion of the vote and the proportion received by each
candidate.

Difference of proportions z-tests were used on each hypothetical race

to determine if the predicted proportion was different from the actual

proportion.

H2: The important variables used to discriminate between two
positive candidates in the one-shot election are the
same as the important variables used to discriminate
between two positive candidates after the campaign pro-
cess.

H3: The important variables used to discriminate between two
negative candidates in the one-shot election are the
same as the important variables used to discriminate
between two negative candidates after the campaign pro-
cess.

H4: The important variables used to discriminate between two
ambivalent candidates in the one-shot election are the same
as the important variables used to discriminate between
two ambivalent candidates after the campaign process.

The discriminant analysis ranks variables according to their im-

portance in discriminating between groups. In each case a Spearman's

r rank test was run between the one-shot group and the campaign pro-

cess group.

HS: Neither issues or personality variables alone in the
race with two positive candidates will discriminate
better than a combination of the two variables.









H6: Neither issues or personality variables alone in the
race with two negative candidates will discriminate bet-
ter than a combination of the two variables.

H7: Neither issues or personality variables alone in the
race with two ambivalent candidates will discriminate
better than a combination of the two variables.

A discriminant analysis was performed on the posttest data from

each of these situations. The model which used the fewest number of

variables for maximum discrimination was used to determine significant

variables.7 Discriminant analyses using just issue and just personality

traits were computed for each campaign. The number of misclassifica-

tions using equal number of variables was examined in order to determine

the best model.

H : There is no difference in the degree of an attitude
before and after the campaign process.

A mean score on the attitudes for each campaign model was computed

using the pretest and the posttest data. For the variables of the

selected candidate, it was determined If the responses of the group had

moved toward "strongly agree," "highly desirable," or "very important."

For the traits of the rejected candidates, it was determined if the

mean responses of the group had moved toward "strongly disagree,"

"highly undesirable," or "very unimportant."















Notes

1. See Osgood, et. al., and David L. Hamilton and Leroy J. Huffan,
"Generality of Impression Formation Process for Evaluative and Non-
evaluative Judgments," Journal of Personality and Social Psychology,
Vol. 20, (1971), p. 200-207.
2. It was felt that the college population would be an adequate
sample for the simulation since the voting population tends to be
"educated, wealthy, employed, white, and males." Robert S. Erikson and
Norman R. Luttbeg, American Public Ooinion: Its Origins, Content, and
Impact, (John Wiley and Sons, New York, 1973), p. 8.
3. Napolltan, p. 120.
4. Fred N. Kerlinger, Foundations of Behavioral Research, (Holt,
Rinehart, and Winston, Inc., New York, 1964), p. 299.
5. Theodore H. White, The Making of the President 1972, (Atheneum,
New York, 1973), Chapters 4 and 8.
6. Swanson, p. 140.
7. As In linear regression, the fewer the variables, the more
reliable the model. See Hubert M. Blalock, Social Statistics, (McGraw-
Hill, New York, 1972), p. 413-415.















CHAPTER III


RESULTS

In this chapter, the variables with their valence computed from

the offset data are listed in Table 1. Next, the predictor models

computed from a discriminant analysis of the offset data are presented

in Table 2.

To test hypothesis 1 that the proportion of the predicted votes

and the actual votes are equal, Table 3 shows the two proportions

and the corresponding z scores. To further explore the accuracy of

the prediction models, the number of correct and incorrect classi-

fications for each case are presented in Table 4.

To determine the accuracy of hypotheses 2, 3, and 4 that the

variables important to the discriminant analysis for the offset data

are the same variables important for discrimination of the posttest

data, Spearman's association of rank test for each of the three

choice situations are presented in Table 5.

For testing hypotheses 5, 6, and 7, which predicted that the model

using both issues and personality traits would be better than using

models having only one of the variables, the results of the three

posttest discriminant analyses for each election are presented in

Table 6.

To examine whether or not there was a change in the degree of

attitudes before and after the campaign simulation, hypothesis 8,









Tables 7-16 report the results of the pretest and posttest discrim-

inant analysis for each case. Included for each election is a table

reporting the comparison of discriminant ability found in the pretest

and posttest data, the discriminant analysis table for the pretest,

and a discriminant analysis table for the posttest.

The results of the t-test for each variable for the control group

is reported in Table 17.


Variables

The offset data indicated that there were five issue statements

with which the subjects agreed, and five personality traits which were

seen as desirable.

Table 1 Issue and Trait Valence From the Offset Data

Variable Valence

Issue

1. Presidential Impeachment +
2. The U.S. should not sell arms to Israel
3. The oil companies are trying to stop ecology +
4. Presidents should be limited to one term
5. Price controls should be reinstated +
6. The ERA should be supported +
7. The $750 per child income tax deduction
should be eliminated
8. Presidential and Congressional elections
should be publicly financed +

Personality Traits

1. inconsistent
2. unagressive
3. follower +
4. progressive +
5. talker
6. dishonest
7. closeminded
8. good natured +
9. aggressive +
10. adventurous +
11. energetic +
(+) = agree or desirable (-) = disagree or undesirable








Predictors

From the offset group, maximum discrimination was reached using

four variables for the ambivalent case and five variables for the posi-

tive and negative cases. The discrimination produced the prediction

function in Table 2.

Table 2 Prediction Functions from the Offset Data and Variables Used
in the Functions with DI indicating Agreement or Desirable
Dimension and D2 indicating Importance Dimension

Ambivalent (candidate A vs. candidate P)

Classify A if 7.04xl 5.05x2 + 4.03x3 + 6.28x 35.36 > 5.07xI 1.74x2
+ 3.11x + 5.23x4 23.05, otherwise classify where
X1 = dishonest (D )
x2 = closeminded (Dj)
x3 = support for the ERA (Dl)
x4 = $750 per child income tax deduction (D2)


Positive (candidate K vs. candidate V)

Classify K if 3.24xl + 3.34x2 + 0.60x3 + 0.54x4 + 8.26x5 44.75 >
3.56xI + 4.02x2 + 0.9x3 + 0.27x4 + 7.28x5 40.95, otherwise classify
V, where
x1 = follower (D )
x2 = aggressive (D2)
x3 = Presidential Impeachment (D1)
x4 = public financing of elections (Di)
x5 = public financing of elections (02)

Negative_(candidate T vs. candidate R)

Classify T if 10.54x 54x + 1.82x + 9.11x + 0.09x 61.09 >
9.29xI 1.09x2 + 1.42x3 + 8.5x4 + 1. 6x5 51 15, otherwise classify
R, where
x, = unaggressive (D )
x2 = impeachment (Di)
x3 = no arms to Israel (D0)
x4 = oil companies stopping ecology (D2)
x5 = $750 per child income tax deduction (02)

Using these models, the pretest data from the campaign group was

applied, and vote percentages were predicted. These percentages were

compared to the percentage of votes actually received after the campaign









process (Table 3). The number of misclassifications for each race was

also computed (Table 4).

Association of rank tests showed that there was a significant

correlation between the ranking of the variables for discrimination in

the ambivalent race when comparing the offset group to the experimental

posttest, but no significant correlation in the positive or negative

cases (Table 5).

Table 3 Comparison of Predicted and Actual Percentages Using Pretest
Predictions and Actual Vote Count from the Experimental Group

Predicted Actual z

candidate A 48.3% 48.3% 0.00
candidate P 51.7% 51.7% 0.00

candidate K 14.5% 43.5% 6.59*
candidate V 85.5% 56.5% 6.59*

candidate T 8.0% 43.5% 10.44*
candidate R 92.0% 56.5% 10.44*
significant at the .05 level

Table 4 Misclassification of Predictions Using the Pretest Data

Data Classification

candidate A candidate P
Voter candidate A 16 14
.Group
candidate P 18 14

-._ _I-_ _- P^S.-- -- --,1. :10
Data Classification

candidate K candidate P
Voter candidate K 2 24
Group
candidate V 7 29

--- -- --- -- -^ta .^^S'J'^-'f- --
Data Classification

candidate T candidate R
Voter candidate T 4 23
Group
candidate R 2 33






33


Table 5 Association of Variable Rank for Offset Data and Experimental
Posttest Data Using Spearman's r

Election r

Ambivalent .55*
Positive -.07
Negative -.05

* significant correlation at the .05 level


Issue vs. Personality

To determine if issue variables or personality variables could

discriminate better than a combination of the two variables, the number

of misclassifications from a discriminant analysis using just issue

variables or just personality variables was compared to a model using

a combination of the two variables (Table 6). In each case, the

posttest model was used because of its superior ability to discriminate.

The issue only and personality only models used for comparison had an

equal number of factors as the combination model in order to keep

reliability factors equal.

Table 6 Comparison of Issue, Personality, and Combination Discriminant
Models Using Posttest Data

Election Number of Classification Classification Percent
Variables Correct Incorrect Droo


Ambivalent

combination 4 44 18 maximum
issue 4 39 23 8.1%
personality 4 44 18 0.0%

Positive

combination 3 47 15 maximum
issue 3 44 18 4.8%
personality 3 43 19 6.5%

Negative_

combination 5 44 18 maximum
issue 5 40 22 6.5%
personality 5 40 22 6.5-









In all three choice situations, maximum discrimination was obtained

when using a combination of the two variables. In the ambivalent case,

however, the use of personality variables alone discriminated equally

as well as using a combination of issues and personality variables. In

the positive and negative condition, a combination of the two variables

discriminated better than using just issue or just personality varia-

bles.


Attitude Change

In order to determine the effect of the reinforcement of attitudes,

the discriminant analysis from the pretest and the posttest were examined

(Tables 7-16). Once again to have equivalent reliability models, the

model which gave maximum discrimination between the voters in the

posttest was compared to a model with an equal number of variables from

the pretest.

A variable was considered to have significant change if it was not

important for the pretest, but became important for discrimination in

the posttest. Traditional test of change, such as t-test, was not

applicable for the interpretation of the data for two reasons. First,

the important factor was not only the movement of one group's mean,

but the relationship of two groups' means in terms of distance between

them. For example, on one variable, the change in a mean score for

group A alone and the change in a mean score for group B alone may not

reach a significant t level. However, if the changes increased the

difference in terms of the distance between the means so that the

variable could be used to discriminate between the groups, then the

change is significant. Secondly, although the means may or may not be









independently significant, in terms of the F score produced by the

analysis, if their contribution was important to the discrimination

model, the changes in means were significant.

Table 7 Classifications from the Discriminant Analysis Pretest
and Posttest

Two Ambivalent Candidates (candidate A vs. candidate P)

Number of Classification Classification
Variables Correct Incorrect

Pretest 4 36 26

Posttest 4 44 18


Table 8 Important Variables from the Pretest Discriminant Analysis
and Means for Candidate Voter Groups

Variable Candidate Mean A Mean P
and Valence

energetic (D1) P + 5.76 5.96
oil companies
and ecology (D1) A + 5.53 5.25
oil companies
and ecology (D2) A + 5.50 5.53
$750 tax deduction (D1) A + 4.93 5.56

(D1 = agreement or desirable dimension
(D2) = importance dimension

Table 9 Important Variables from the Posttest Discrimination Analysis
And Pretest And Posttest Means for Candidate Voter Groups

Variable Candidate Pretest Posttest Pretest Posttest
and Valence Mean A Mean A Mean P Mean P

dishonest (D,) A 1.76 2.06 1.68 1.34
energetic (Di) P + 5.76 5.60 5.96 5.96
oil companies
and ecology (DI) A + 5.53 5.70 5.25 5.06
supporting
the ERA (02) A + 5.40 5.76 5.59 5.37


The posttest model included two variables not used in the pretest

model. The first variable, dishonest, increased in desirability for

candidate A's voters and decreased in desirability for candidate P's









voters. The other new variable, support for the ERA, had increased

agreement for candidate A's voters and decreased in the level of agree-

ment for candidate P's voters. The other two variables held in their

relationship, with voters who selected the candidate having the trait

apparently seeing the variable as more positive than the other group.

It was interesting to note that although these variables were important

to the discrimination on the pretest, the distance between the groups'

means increased on the posttest.

Table 10 Classifications from the Discriminant Analysis Pretest and
Posttest

Two Positive Candidates Jcandidate K vs. candidate Vi

Number of Classification Classification
Variables Correct Incorrect

Pretest 3 42 20

Posttest 3 46 16



Table 11 Important Variables from the Pretest Discriminant Analysis
And Means for Candidate Voter Groups

Variable Candidate Mean K Mean V
and Valence

follower (D2) K + 4.85 5.22
adventurous (D1) V + 4.11 4.54
reinstatement of
price controls (DI) V + 4.07 4.85


Table 12 Important Variables from the Posttest Discriminant Analysis
And Pretest and Posttest Means for Candidate Voter Groups

Variable Candidate Pretest Posttest Pretest Posttest
and Valence Mean K Mean K Mean V Mean V

good natured (D1) K + 4.81 4.96 4.88 4.34
aggressive (D2) V + 5.22 5.44 5.68 5.88
impeachment (D2) K + 6.07 6.25 5.48 5.74









None of the three variables used in the posttest discrimination

were important in the pretest discrimination. The first variable, good

natured, increased in desirability for candidate K's voters and decreased

in desirability for candidate V's voters, thus increasing the distance

between them. It was interesting to note that on the pretest, the

trait apparently was seen as more desirable by candidate V's voters than

by candidate K's voters.

The other two variables, aggressive and impeachment, were more

important for voters who selected the candidate having the trait or

advocating the issue. Although the distance between the means on

aggressive did not increase, and the means for impeachment increased

only slightly, these two variables became important to the discrimination

on the posttest. The reason for this occurrence was that the within

group variance of both groups decreased, thus forming distinguishable

clusters for the discrimination. The important implication fro. this

occurrence in relation to social judgment theory will be discussed in

Chapter IV.

Table 13 Within Group Variance for Aggressive and Impeachment: Pretest
and Posttest Comparison

Variable Pretest Posttest Pretest Posttest
Variance K Variance K Variance V Variance V

aggressive 1.25 .94 1.14 .56
impeachment 1.26 .66 1.87 1.12









Table 14 Classification from the Discriminant Analysis Pretest and
Posttest


Two Negative Candidates (candidate T vs. candidate R)


Classification
Correct
50


Classification
Incorrect
12


Table 15 Important Variables from Pretest Discriminant Analysis And
Means for Candidate Voter Groups


Variable Candidate
and Valence

unaggressive (D1) R -
talker (DI) T -
no arms to
Israel (DI) T -
reinstatement of
price controls (DI) R -
reinstatement of
price controls (02) R -


Mean T


Mean R


2.51
3.48

4.55

5.22

4.96


3.05
4.17

3.60

5.74

4.17


Table 16 Important Variables from the Posttest Discriminant Analysis
And Pretest And Posttest Means For Candidate Voter Groups

Variable Candidate Pretest Posttest Pretest Posttest
and Valence Mean T Mean T Mean R Mean R

inconsistent (D2) R 5.92 5.77 5.91 5.52
talker (D2) T 4.70 4.77 5.31 5.28
no arms to
Israel (DI) T 4.55 4.59 3.60 3.65
reinstatement of
price controls (D2)R 4.96 5.03 4.17 4.34
$750 tax
deduction (D2) R 5.18 5.51 5.40 5.28


For the variables inconsistent, talker, and the $750 tax deduction,

the importance of these negative variables lessened for the groups

which selected the candidate and increased for the group which rejected

the candidate. The other two variables of no arms to Israel and price


Pretest

Posttest


Number of
Variables
5

5






39

controls changed only slightly in the distance which was previously

significant for pretest discrimination. On both issues, voters who

selected the candidate taking this position apparently saw the stand

as less negative than voters who rejected the candidate.


Control Group

There were no significant changes on any variable in the control

group condition (Table 17).

Table 17 Control Group t's for Issue and Personality Trait Variables

Variable

Issue t

1. Presidential impeachment (Di) 1.65
2. Presidential Impeachment (D2) 0.79
3. The U.S. should not sell arms to Israel (Di) 0.74
4. The U.S. should not sell arms to Israel (D2) 0.89
5. The oil companies are trying to stop ecology (Di) 0.13
6. The oil companies are trying to stop ecology (0 2 1.77
7. Presidents should be limited to one term (D ) 0.17
8. Presidents should be limited to one term (D2) 0.64
9. Price controls should be reinstated (D ) 0.05
10. Price controls should be reinstated (D2; 0.27
11. The ERA should be supported (D0) 0.01
12. The ERA should be supported (D2) 1.79
13. The $750 per child income Lax deduction
should be eliminated (Di) 0.79
14. The $750 per child income tax deduction
should be eliminated (D2) 0.32
15. Presidential and Congressional elections
should be publicly financed (Di) 0.51
16. Presidential and Congressional elections
should be publicly financed (D2) 0.64








Table 17 (continued)

Personality ______ t -

1. inconsistent (D1) 1.06
2. inconsistent (D2) 0.59
3. unaggressive (DI) 0.05
4. unaggressive (D2) 1.13
5. follower (D ) 0.21
6. follower (D2) 1.19
7. progressive (DI) 1.09
8. progressive (D2) 1.47
9. talker (DI) 0.33
10. talker (D2) 0.80
11. dishonest (DI) 1.14
12. dishonest (D ) 0.03
13. closeminded (DI) 0.95
14. closeminded (D2) 0.03
15. good natured (Dl) 0.31
16. good natured (D2) 0.03
17. aggressive (DI) 1.41
18. aggressive (D2) 1.10
19. adventurous (D) 0.54
20. adventurous (D2) 0.80
21. energetic (DI) 1.18
22. energetic (D2) 0.22















CHAPTER IV


CONCLUSIONS AND DISCUSSION

In this chapter, there are five sections. First, the results of

the testing of the hypotheses were reviewed. Next, there was an

evaluation of the use of simulation. There were three pieces of

evidence which were discussed that indicate that the use of simulation

for the study of political communication is a reasonable.methodology.

Also, implications from the use of simulation about the nature of

political campaigns were examined.

Third, there was an examination of issues and personality traits

as variables influencing voting behavior. The indications that the

variables may function differently in ambivalent, positive, and negative

choice situations was discussed. Also, a possible problem in the

structuring of the ambivalent candidates was discussed and implications

about the use of issues and personality traits were drawn.

Fourth, support for the social judgment theory in terms of attitude

change was explored. The fact that every variable followed one of the

predictions of social judgment theory and that there was no evidence

of counter-attitudinal behavior were examined in terms of the connection

between political attitudes and voting behavior.

Finally, implications for future research were discussed for three

possible new directions. First, the indications for future theory

building and model building were examined. Secondly, the need for an

examination of the communication process was discussed. Finally,









implications for field research were examined.


Conclusions

Hypothesis 1, which predicted that there would be no significant

difference between the predicted proportion of votes and the actual

proportion of votes for the hypothetical Congressional candidates, was

not rejected for the ambivalent election, but was rejected for the

positive and negative elections. The accuracy in the ambivalent case,

however, was found to be a function of a number of misclassifications

for each group cancelling each others' effect.

Hypotheses 2, 3, and 4 predicted that the important.variables for

discrimination of the offset voting data would be the same variables

important for the discrimination of the posttest data of the simulation

group. For the positive and negative cases, hypotheses 2 and 3, respec-

tively, the hypotheses were rejected. In the ambivalent case, the

hypothesis was not rejected, therefore indicating an association

between the two rankings.

Hypotheses 5 and 6, which predicted that models using both issues

and personality traits could discriminate between voter groups better

than models using only one of the variables for the positive and nega-

tive cases, were not rejected. Hypothesis 7, however, which made the

same prediction for the ambivalent case, was rejected since an analysis

using only personality traits had equal ability to discriminate between

groups of voters as a model using a combination of issues and person-

ality traits.

Hypothesis 8, which predicted that there would be no difference

in the degrees of attitudes before and after the campaign process, was









rejected because in every condition when means for each voter group on

the variables important in the posttest discriminant analysis were

examined, it was found that the variables changed as predicted by the

social judgment theory. The nature and implications of these changes

will be discussed later in the chapter.


Discussion

The Simulation and Its Effects

The results from this study suggested that although the predictions

of voting behavior were inaccurate, the experimental treatment seemingly

had an effect on the subjects. There were several indications which

warrant this conclusion.

First, although the predicted percentages for the ambivalent case

were exact, when the number of misclassifications resulting from the

use of the predictor model were computed, it was seen that the accuracy

of the model was partially the result of the fact that two misclass-

ifications cancelled each other in the computation of percentages. In

the positive case and negative case, the predicted winners for the

election were correct. However, the landslide margin that was pre-

dicted was not near the actual percentage of votes received. The fact

that the subjects' initial attitudes were poor predictors of the

subjects' responses to the set of those attitudes (the candidates')

after the simulation apparently indicates that the process altered

the initial responses.

Another indication that the experimental campaign had an effect

upon the subjects was the difference of ranking of the variables

between the offset group and the experimental posttest. Although there









was a significant correlation between the overall ranking of variables

in the ambivalent case, there was no significant correlation in the

positive or negative cases.

One possible explanation for these results would be the nature of

the decision-making situation. In the positive or negative cases, the

voter would have to engage in more cognitive reorganization to ration-

alize the choice as opposed to the ambivalent case where the voter

could find both positive and negative elements in the situation. This

explanation is parallel to explanations of dissonance relief in decision-

making contexts. As Kiesler, Collins and Miller explain: "The less

attractive the chosen alternative, the greater the dissonance. The

more attractive the unchosen alternative, the greater the dissonance."

The ambivalent candidates would be between these alternatives, and

therefore, create less dissonance.

The final piece of evidence which indicated that the experimental

communication treatment affected the subjects was the comparison of

variables in the discriminant analyses of the pretest and posttest in

the experimental condition. In every case, some new variables which

were not important for the discrimination using the pretest data became

important to the discrimination when using the posttest data. This

fact indicated that during the campaign process, a reordering of varia-

bles took place. If this reordering was due to the events outside the

experiment, changes should have been observed in the control group.

Since these changes were not seen in the control group, the reordering

was apparently the result of the subjects' experiencing the simulated

campaign process and decision-making task.

The indications that the simulated campaign process affected the









subjects has several important implications for studies of political

communication in a field setting. First, since the campaign process

may cause a restructuring of variables involved in the voters' decision,

basing the structure of the campaign communication on research gathered

well in advance of a campaign may not be the best strategy. Secondly,

it would be important in a field setting to understand the voters'

view of the type of choice they have. That is, if the voters see the

candidates as ambivalent, then it may be easier to locate important

variables than if all the candidates are seen as positive or negative.

Finally, when attempting to understand the development of voters'

decisions in a field setting, it would be important to measure voters'

attitudes a number of times during the co-mmunication process. If the

communication strategy is to be evaluated and adjusted during a campaign,

then it would be imperative to locate new, important variables as they

occur. In the practical sense of politics, understanding the changes

resulting from the communication process after the election has limited

applicability. In terms of research of political communication, such

intervening measures, both in field settings and controlled simulations,

may clarify the development of the decision-making process.


Issues and Personality Traits As Variables

In the examination of issues and personality traits, the results

from the positive and negative cases were similar, but these situations

were different from the ambivalent case. In discriminating between voters

in the choice between two positive candidates and between voters in the

choice between two negative candidates, using a combination of the

issues and personality variables produced a better model than using








only one of the variables.

The fact that neither variable alone could discriminate between

voters better than a combination of the variables seems to indicate

that these variables operate situationally. That is, at least in the

choice between two positive candidates and the choice between two

negative candidates, any variable cannot be assumed to be more important

than another variable because it is an issue or personality trait.

There was apparently another factor, such as ego-involvement which

determines the influence of a variable in a voter's decision.

In the ambivalent choice situation, using issues alone was just as

accurate as using a combination of issues and personality traits, but

issues alone was not as good as a combination model. This result may

have been due to the structuring of the candidates for the simulation

and the attempts to reinforce the attitudes with communication. The

two candidates were structured with one candidate being positive on

issues and negative on personality traits, and the other candidate

being constructed oppositely. If the reinforcement caused attitudinal

change in the predicted direction, then the voters who selected the

candidate with positive issues should show more extreme attitude pos-

itions about the issues. The same should hold true for attitudes about

personality traits for the voters who selected the candidate with pos-

itive personality traits.









Table 18 Means for Agreement and Desirability Dimensions for the
Issue-Only Model and Personality Traits-Only Model in the
Ambivalent Case with Candidate A Being Positive on Issues
and Candidate P Being Positive on Personality Traits
Issue Mean A Mean P
li ------------- ------------

oil companies and ecology 5.70 5.06
limitation of President to one term 3.13 3.46
support for the ERA 5.76 5.37
$750 per child tax deduction 3.23 3.71

Personality Traits
f^^S1.^. I'9!- -------------------'------

progressive 5.73 5.90
dishonest 2.06 1.34
closeminded 2.63 2.00
energetic 5.60 5.96


From an examination of the posttest means of the "agreement" and

"desirability" dimensions, it seems that the explanation is valid. The

group who voted for Candidate A, who was positive on issues, showed more

extreme positions in their opinions about issues. The group who voted

for Candidate P, who was positive on personality traits, showed more

extreme positions in their opinions about personality traits.

It was interesting, however, that the personality traits became

distinct enough to discriminate equally as well as using a combination

of the two variables. This finding may indicate a difference in the

level of reinforcement through communication which can be achieved for

these two variables. For example, it may be easier to reinforce attitudes

about personality traits since the judgments would not be influenced by

the level of information about current events, but would be dependent

upon a belief structure which has been formed through socialization.

Support for the Social Judgment Theory

The concept that responses to variables could be predicted in terms
of social judgment theory was substantiated in three ways. First, the
means for the variables which were important for pretest discrimination








and held in this importance for the posttest discrimination indicated

that the group who selected a candidate having a certain trait or

position saw the trait as more positive, or at least less negative,

than the group who rejected the candidate.

This fact, when interpreted in terms of latitudes of acceptance

and rejection, suggested that the traits of the selected candidate

were closer to the "most acceptable" position, and the traits of the

rejected candidate were closer to the "most unacceptable" position.

An example of this type of occurrence was the variable of sale of arms

to Israel in the negative case (Tables 15 and 16). Although claiming

that no change in the relationship was significant seems contradictory,

this concept is important when the fact that many other variables

within the multivariate situation were changing, yet this important

relationship was stable.

The second type of variable functioning which supported interpre-

tation by social judgment theory was the direction of changes in mean

scores of variables which were not important in the pretest discrim-

ination, but became important to the posttest discrimination. The

changes in these mean scores indicated attitudinal change was in the

direction that would be predicted by assimilation-contrast.

In all three choice situations, the new variables' means for the

group who selected the candidate associated with the variable increased

in a positive direction. Also, the new variables' means associated

with the rejected candidates decreased in a negative direction. In all

groups, there was no evidence of counter-attitudinal behavior, i.e.,

for every group,- the variables associated with the selected candidate

were apparently seen as more acceptable (or less objectionable) than









the variables associated with the rejected candidate. Examples of this

concept are support for the ERA (Table 9), in the ambivalent case, good

natured (Table 12) in the positive case, and inconsistent (Table 16) in

the negative case.

The finding of consistent behavior and attitude responses is

particularly important in the light of previous research attempting to

relate attitudes and behavior. In many previous field studies, dis-
2
crepancy was found between reported attitudes and behavior. The fact

that the two were consistent tends to support the previous premise

that voting behavior may be distinct from other types of behavior in

the sense that someone else is selected to carry out the actual action

described by the attitude.

Finally, social judgment theory predicts that the reinforcement

of an attitude in the latitude of acceptance can make the position

"most acceptable." The concept was apparent when variables were not

important in the pretest but became important in the posttest, although

the distance between means remained relatively stable. The variables

became important because within-group variances became smaller. Since

the mean increased and the variance became smaller, it seemed that the

group's view of the variable clustered at a point near the extreme or

"most acceptable" position. Examples of the occurrence were seen in

the positive case on the variables of aggressive and impeachment

(Table 12).

Since every important variable followed one of the predictions of

social judgment theory, it seems that this theory offers a sound

interpretation of voting behavior based on these two variables. There

is one other piece of evidence which must be examined, however, before









accepting this generalization. If the variables, through stabilization

in latitudes of acceptance and rejection, assimilation-contrast, and

clustering toward the "most acceptable" point, were consistent, then

the groups who chose each candidate should become more distinctive

for the discriminant analysis.

This concept can be seen through a comparison of the number of

misclassifications of the pretest data and the posttest data. In the

ambivalent case and the positive case, the posttest data were superior

for discrimination when compared to the pretest data.

In the negative case, the posttest discrimination was not as good

as the pretest discrimination. Although the mean scores functioned in

the predicted manner, the poorer discrimination indicated a wider

variance in scores which clouded the discrimination. This fact indi-

cated that attempting to reinforce negative attitudes in an attack

manner with no supportative reinforcement of positive attitudes may

cause a shift toward a non-committal position in some cases. Such an

implication would have significant meaning in terms of political commun-

ication strategy. It may not be as effective to use only attacks upon

an opponent as it would be to use only supportative, or supportative

and attacks during a campaign.

Despite this one question raised in the negative condition, the

examination of political communication variables with the predictions

of social judgment theory appears to be a reasonable approach considering

the functioning of the variables in this study. It should also be noted

that there was support for the view that political attitudes are multi-

dimensional since the changes seemed to occur on both the "agreement"

or "desirable" dimensions and the "importance" dimension with approxi-








mately equal frequency.

If a social judgment theory approach to political communication

is relevant, it explains why field studies have located different

variables as significant in different elections. Attempts to debate

the relative influence of issues or images may be a meaningless

discussion since the functioning of variables in these categories may

be dependent upon the voters' preconceived acceptance or rejection of

items regardless of the category.


Implications for Future Research

This study indicated that it is both feasible and possible to use

simulation as a methodology for the study of political communication.

Through the use of this somewhat ignored methodology, future research

can make two important contributions to the study of political communi-

cation. First, through continued use, refinements of methods can be

gained and problems overcome. For example, in this study, the structure

of the ambivalent candidates may have produced results which were easily

misinterpretable when comparing an issue model, a personality model, and

a model using a combination of these variables. As these problems are

discovered, necessary adjustments can be made.

Secondly, the use of voter simulation gives a way for findings to

be confirmed through repetition and allows for any number of variables

to be studied. If methodological mistakes are discovered, then the use

of simulation allows for correction of the mistakes and the repetition

of the study. In a field setting during an actual election, this

luxury does not exist. If mistakes are made the data are lost.

The simulation also allows future studies of political communication









to add or subtract any number or type of variables. Through the system-

atic controlling of the variables within the model, the validity of

using social judgment theory interpretations can be tested. From the

results of this study, for example, the next logical extension should

be the adding of a new communication variable to determine if the

functioning of issues, personality traits, and the new variable add

support to the social judgment theory interpretation, or support

another alternative. Possible new variables which could be used to

expand the model would be sex of the candidate, endorsement by others

who have high or low credibility with the electorate, peer agreement,

party affiliation, or contact with the candidate.

Another direction that future research should take is the dissecting

of the political communication process which was used in the study.

Although the social judgment interpretation appears reasonable for

interpretation of voting behavior, to gain in-depth understanding of

the communication process, isolated examination of parts of the communi-

cation process needs to be explored. This study strongly suggested,

for example, that an attack structured message may function differently

than a supportive message. Likewise, the relative strengths and weak-

nesses of different channels' contributions to the reinforcement of

attitudes needs to be explored.

The attitude responses to reinforcement also suggest a new

direction for communication study. With the amount of change which

was exhibited in all three situations, the power of persuasion through

the reinforcement of attitudes seems to warrant further investigation.

The concept of reinforcement of attitudes to produce change may have

significant meaning in context other than political communication. The









changes produced in the study also support the basic contention of

Swanson that dealing in acceptable limits does not mean stagnation,

but merely sets parameters in which the communicator must work.3

Finally, future research must take the major concept in this

study, the voting response to a campaign is a function of social

judgment and ego involvement, and test its validity in a field setting.

One important investigation, for example, would be to test the validity

of the interpretation for campaigns designed for different public

offices than Congressional. The confirmation or rejection of the theory

with any new implications can then be returned to the controlled situa-

tion for further examination. If this circular type research procedure

can be accomplished, the constructive theory building in political

communication will be sound and meaningful, and the probability for

understanding increased instead of confused.















Notes

1. Kiesler, Collins and Miller, p. 201
2. For example, see R.D. Minard, "Race Relationships in the
Pocahontas Coal Field, Journal of Social Issues, (1952), p. 29-44,
and R.T. LaPiere, "Attitudes vs. Action," Social Forces, (1934),
p. 230-237.
3. Swanson, p. 133.















APPENDIX A


DESCRIPTION OF THE CANDIDATES

Race 1 Candidate A vs. Candidate P

Candidate A has been a strong supporter of the Equal Rights Amend-
ment because he feels that there has been sex discrimination in many
job fields. He has also attacked the big oil companies for taking
advantage of the energy crisis to reverse policies proposed by ecology
groups. He has been opposed to limiting any President to one term in
office.
While serving in office, he has refused to support any legislation
which was not introduced by a member of his political party. Recently,
a citizens group investigating a number of candidates found evidence
that some of his campaign funds were coming from organized crime.

Candidate P favors the limitation of the President to one term
because he feels that this action will help end the use of government
employees to run political campaigns. He is also in favor of a bill
which would take away the $750 per child deduction from Federal income
tax. Candidate P has opposed the Equal Rights Amendment because he
feels that it is unnecessary.
While in business, Candidate P managed three companies at one
time. He also found time to work as a "big brother" to underprivileged
children and head the United Way fund raising campaign. He also helped
raise funds to start an experimental school that would use new tech-
niques for the teaching of the mentally retarded.

Race 2 Candidate K vs. Candidate V

Candidate K believes that the President should be impeached be-
cause he has lost credibility with the American people. He also pro-
poses election reforms by the public financing of the Predisential and
Congressional campaigns. He takes a lot of kidding about being bald,
but he just smiles and says "I'm not bald, other men are just too
hairy." He has never introduced new legislation, but he has actively
supported many bills proposed by other congressmen.

Candidate V wants to return price controls to food in order to
slow down inflation. He opposes the limitation of the President to one
term. Candidate V started his own business and made it into one of the
largest companies in the state. He has never proposed a bill in the
legislature that has been defeated, although some of them were quite
controversial. Candidate V spends his free time camping and traveling
by canoe down rivers all over the U.S.









Race 3 Candidate T vs. Candidate R

Candidate T is opposed to the U.S. selling arms to Israel as he
feels this constitutes an unnecessary military involvement in the area.
He has supported the big oil companies because he feels much of the
ecology considerations which they are forced to have are really unneces-
sary and hurt the economy. Candidate T feels that thus far no evidence
has been presented which would warrant the impeachment of the President.
In his own community, Candidate T has proposed many citizen commit-
tees to solve local problems. However, he has never actively partici-
pated in the organization of these groups.

While serving as a state legislator, Candidate R proposed much
legislation. When they met with opposition, however, he did not push
for their passage. Because he feels that overpopulation is the real
cause of the energy crisis, he favors a proposal that the $750 per child
Federal income tax deduction be ended. He also opposes price controls
on foods as he sees this action as government interference with free
enterprise. Candidate R changes his mind often, however. Sometimes on
an issue stand he has previously taken.















APPENDIX B


TV ADS

I. Picture- little boy and little girl

Voice: Do either of these children look handicapped? One of them is.
Look close. The child on the left has a handicap from birth because
this child is female. Unless something changes, she will be offered
fewer job opportunities, less money, and fewer promotions. Candidate A
wants to do something about it. He supports the Equal Rights Amendment.
Vote Candidate A for Congressional seat I. Let's not handicap our
nation.

II. Picture- Man asleep

Voice: Some people would rather sit and let the world pass them by.
Candidate P, however, wants to make this land a better place. Although
managing three companies, he still found time to be a "big brother" to
underprivileged children. He also headed the United Way drive in his
community. Vote Candidate P for Congressional seat 1 and get a man who
cares.

III. Picture- White House five

Voice: Mitchell, Dean, Hunt, Erlichman, Halderman; Five reasons why
the President has lost credibility. A national disgrace has been the
main contribution of this administration. With the top five presiden-
tial aides in jail, no wonder the American people have had enough of
this administration. Candidate K favors the impeachment of the
President. We need to restore our house to order. Vote Candidate K
for Congressional seat 2.

IV. Picture- River

Voice: To take care of nature, you have to understand it. Candidate V
spends his free time camping and traveling by canoe down rivers all
over the United States. He understands nature. Vote for Candidate V
for Congressional seat 2 and help preserve what we have left.

V. Picture- smile-frown face

Voice: How does this guy feel? Hard to tell, isn't it. Candidate R
is much like this. He often changes his mind about issue stands he has
previously made. It's hard to tell how he feels on any issue. Do we
need our leaders to be indecisive? Get real leadership, vote Candidate
T for Congressional seat 3.






58


VI. Picture- plaque of soldiers

Voice: Many men have died to protect their freedom. In Israel, the
constant threat of war is a reality. The Russians are giving arms to
the Arabs. Israel wants to buy arms from us. Should we deny these
people the right to protect themselves? Candidate T thinks so. He
wants to forbid the sale of arms to Israel. Don't let this country
turn its back on one of the few democracies left. Vote Candidate R for
Congressional seat 3.















APPENDIX C


RADIO ADS

I. Can you imagine a world without fish, beaches, or birds? Luckily,
some people foresaw our plight and started working to preserve our
environment. But the tide is changing because the oil companies have
convinced the government that immediate needs are more important than
future generations. Candidate A wants to make sure that those who come
after us will have something to inherit. He has pledged to fight for
the preservation of ecological programs despite the wishes of big oil
companies. Help keep our earth alive. Vote Candidate A for Congres-
sional seat 1.

II. If America is to be strong in the future, education will play a
vital role. Not just the education of a few, but the education of all
people. Candidate P helped raise funds to start an experimental school
for the teaching of the mentally retarded. Get progressive action.
Vote for Candidate P for Congressional seat 1.

ill. You've heard that money can buy anything. Recently, that sure
has seemed true in national politics. To keep a small group of people
from influencing elections, Candidate K believes that all Presidential
and Congressional elections should be publicly financed. Give the
power back to the people. Vote for Candidate K for Congressional seat 2.

IV. Candidate V started his own business and made it into one of the
largest companies in the state. He has been elected to the state legis-
lature four times previously. He has never proposed a bill in the
legislature that has been defeated, although some of them were quite
controversial. Get a man of action. Vote for Candidate V for Congres-
sional seat 2.

V. Talk is cheap. Candidate T did a lot of talking. He proposed many
citizen committees to solve local problems. However, he never actively
participated in the organization of these groups. Don't we have enough
talkers in Congress? Vote for Candidate R for Congressional seat 3.
Get action, not words.

VI. Should the government try to control your family's lives? Candi-
date R seems to think so. He favors the repeal of the $750 per child
income tax deduction. Government should not control people, people
should control government. Keep your voice heard. Vote for Candidate T
for Congressional seat 3.















APPENDIX D


THE CITIZENS' BUREAU REPORT ON
THREE CONGRESSIONAL CAMPAIGNS

(Compiled by the examination of
campaign literature, independent
observation, and analysis of pre-
vious records.)

Congressional seat 1 Candidate A vs. Candidate P

Candidate A has opposed the limitation of the President to one
term in office. He feels if a person does a good job for four years,
the people have the right to give him another four years in office.
Candidate A has been critized by both the press and his opponents
because he has refused to support any legislation which was not pro-
posed by a member of the same political party that he belongs to.
Recently, a citizens group investigating political candidates claimed
that they found evidence of campaign funds coming from organized crime.

Candidate P disagrees with Candidate A's stand on the limitation
of the President to one term. He feels that a one term limitation
would help stop the use of government crployees, paid by the taxpayer,
in running political campaigns. In his home state, Candidate P has
been a strong opponent of the Equal Rights Amendment. He stated that
there are too many flaws in the ERA, and it is really not needed.
Candidate P also claimed that the major cause of the energy crisis has
been overpopulation. As an incentive for population control, he favors
the elimination of the $750 per child deduction on Federal income tax.

Congressional seat 2 Candidate K vs. Candidate V

Candidate K has enjoyed good rapport with his fellow congressmen
and the press. Although he is often kidded about being bald, he usually
smiles and replies; "I'm not bald, other men are just too hairy."
While in Congress, Candidate K never proposed any original legislation,
but he did support many bills proposed by other congressmen.

Candidate V feels the major problem facing this country is infla-
tion. To help stop inflation, he believes that the government should
continue to control food prices. Candidate V thinks that the proposed
limitation of the President to one term is an over reaction to the
present situation, and therefore, he is opposed to the action.






61


Congressional seat 3 Candidate T vs. Candidate R

Candidate T has spoken out against the impeachment of the President.
He feels that so far, there has been no evidence of activities which
warrant such action. He also feels the relaxation of pollution laws is
necessary in order for the big oil companies to find new sources of
needed energy.


While serving as a state
legislation. He did not push
with opposition. Candidate R
that such action is an attack


legislator, Candidate R proposed
for their passage, however, when
is against food price controls.
on the free enterprise system.


much
they met
He claims
















APPENDIX E


THE CONGRESSIONAL PREVIEW

.(A Report Compiled by UPI
Political Correspondents)

NOTE: The description in this message included all the variables and
Is a duplication of Appendix A.















BIOGRAPHICAL SKETCH

James Travis Kitchens was born May 8, 1949, in Atlanta, Georgia.

He received his early education in Atlanta and in 1967 graduated from

Southwest High School. In 1971, he received his B.S. degree in Speech

Education from the University of Georgia. His M.A. degree was earned

in 1972 at the University of Georgia. In 1972, Mr. Kitchens entered

the University of Florida, and his Ph. D. was conferred in 1974.















I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is
fully adequate, in scopy and quality, as a dissertation for the degree
of Doctor of Philosophy.



Douglas C.. Bock, Chairman
Associate Professor of Speech



I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is
fully adequate, in scope and quality, as a dissertation for the degree
of Doctor of Philosophy.



Donald E. Williams
Professor of Speech



I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is
fully adequate, in scope and quality, as a dissertation for the degree
of Doctor of Philosophy.



Anthony J. ClarK
Associate Professor of Speech



I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is
fully adequate, in scope and quality, as a dissertation for the degree
of Doctor of Philosophy.



Thomas J. refine
Assistant professor of Speech
















I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is
fully adequate, in scope and quality, as a dissertation for the degree
of Doctor of Philosophy.




Mi ael Cornett
A istant Professor of Speech


I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is
fully adequate, in scope and quality, as a dissertation for the degree
of Doctor of Philosophy.




Alan Agresti
Assistant Professor of Statistics


This dissertation was submitted to the Department of Speech in the
College of Arts and Sciences and to the Graduate Council, and was
accepted as partial fulfillment of the requirements for the degree of
Doctor of Philosophy.




Dean, Graduate School


December, 1974










































UNIVERSITY OF FLORIDA

3 1262 I 073IIII32 I18
3 1262 07332 031 8