ABSTRACT

COUNTBRPATTITUDINAL ENCODING EFFECTS
ON MESSAGE STYLE AND PERFORMANCE
by William James McEwen

Experimentation into the research area entitled "counter-attitudinal
encoding," through its fecus on attitudinal shift, has largely neglected
to undertake a thorough analysis of the communication behavior of encoders.
This study therefore addressed itself to an examination of the nature of the
counter-attitudinal encoding process, i.e. the message performance and style
characteristics which are affected by whether or not a writer agrees with the
message output required of him.

Research into the effects of psychological stress on communication per-
formance/suggests that a general decrement in perfbrmance quality will occur
as a result of heightened stress. Assuming that counter-attitudinal encoding
requirements generate some amount of stress, this performance decrement should
be reflected in a slower rate of message output as well as an increasing number
of message errors and changes in counter-attitudinal messages.

Research into stress or motivation effects on communication behavior
also suggests that a generally lower level of message readability will result
from an increase in the stress level under which an encoder is operating.

In addition, available research indicates that an encoder's hesitancy
or lack of certainty is reflected in the language he subsequently uses. Varying

levels of language intensity have been found to be related to the amount of

William James McBwen

polarity or extremity eXpressed in two principal message elements: verbs and
modifiers. It would also appear that certain types of conjunctions reflect

the degree of hesitancy present in a message source. Therefore, assuming that
encoders of belief-discrepant messages eXperience greater hesitancy about what
they are required to write as well as an increase in cognitive stress, specific
message consequences would be expected. Counter-attitudinally produced messages
should contain language elements indicative of lower expressed intensity than
attitudinally-consistent messages.

All experimental subjects wrote one message which argued in accord with
a previously stated opinion and one message which argued in opposition to this
opinion. The order in which these messages were written was contolled for, as
was the evaluation of the message topic (i.e. whether subjects wrote on a
positively or negatively evaluated topic). ,Message style and performance com-
parisons were made between each subject's counter-attitudinal and attitude-con-
sistent messages.

It was found that encoders made somewhat more errors when arguing counter-
attitudinally than they made when writing attitude-consistent messages, although
no difference in message output rate was found.

Encoders also tended to use less intense language when writing counter-
attitudinal messages than they used in their attitude-consistent messages. This
language intensity decrement was most evident in the apparent mood (more
conditional) and voice (more passive) of the verbs used by counter-attitudinal
encoders. No difference was observed in the readability levels of counter-

attitudinal vs. attitude-consistent messages.

William James McEwen

Possible reasons for the incomplete confirmation of the experimental
hypotheses are suggested. Research extensions of the present study and impli-
cations for future research into the probable impact of counter-attitudinally

produced messages on receiver evaluations and exposure probabilities are dis-

cussed.

Accepted by the faculty of the Department of Commmication,
College of Communication Arts, Michigan State University, in partial

fulfillment of the requirements for the Doctor of PhilOSOphy degree.

ire “0% @533“?

 

Guidance Committee :

 

 

 

 

 

COUNTER-ATTITUDIN AL ENCODING EFFECTS

ON MESSAGE STYLE AND PERFORMANCE

By

.“ ..)

n” "
William J? McEwen

A THESIS

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

DOCTOR OF PHILOSOPHY
Department of Connnunication

1969

,‘57'73 0).
7-- 3'07

ACKNOWLEDGMENTS

At this point in one's graduate career, it seeus highly appropriate to
turn around and look back. Having discovered where you are now, it is fitting
to consider how you got there.

The list of individuals who have helped to get me as far as this plateau
seens endless. Unfortunately, the thanks which can be rendered them appear
entirely inadequate.

For their constant but patient commitment to seeing their son develop
whatever potential he had, let me first thank my parents.

For providing a whole new world for discovery, and for aiding me to
uncover bits of it, let me thank both the faculty and the students of the
Department of Communication at Michigan State.

For their suggestions, help and guidance, let me thank my doctoral
committee members: Dr. V. C. Troldahl, Dr. E. P. Bettinghaus,

Dr. G. R. Miller and Dr. A. M. Barch. Host} especially, let me thank my
chairman and advisor, Dr. Bradley S. Greenberg. Whatever skills and com-
petencies I may have developed as an investigator of human symbolic behavior
are due in very large part to the direction provided and the standards demanded
by this one man. I A

For service above and far beyond the call of duty in the statistical
analyses employed in this dissertation, let me warmly thank Dr. Bradley Lashbrook.

For her encouragement, warmth and understanding throughout my graduate
career, and for creating an environment where the pursuit of knowledge is both
rewarding and exciting, let me thank my wife Florrie.

ii

TABLE OF CONTENTS

CHAPTER
I INTRODUCTION . . . . . . . . . . .
II METHODS AND PROCEDURES . . . . . .

Design and Analysis . . . . . .
Procedures . . . . . . . . . . .
Independent Variables . . . . .
Control Variable . . . . . . .
Dependent Variables . . . . . .

III RESULTS . . . . . . . . . . . . .

Performance Variables . . . . .
Message Style Variables . . . .
Combined Analysis . . . . . . .
Sign Test Analysis . . . . . . .
Analysis of Variance Results

Individual Mood and Task Ratings

IV DISCUSSION . . . . . . . . . . .
Methodological Problems . . . .
Theoretical Issues . . . . . . .

Research Implications . . . . .

REFERENCES 0 O O O O O O O O O O O O O 0

1k

14
15
17
18
19

25

27
31
37
39
#2
H2

#5
#8
S2
53

59

7.

8.

9.

10.

11.

12.

13.

LIST OF TABLES

Mean Encoding Rate and Time Spent Encoding
Message Errors . . . . . . . . . . . . .
Message Error Rate . . . . . . . . . . . .
Readability Measures . . . . . . . . . . .
Modifier Polarity Proportion . . . . . . .
Verb Form Polarization Proportion . . . .
Verb Voice Proportion . . . . . . . . . .
Verb Tense Proportion . . . . . . . . . .
Ambivalence Construction Proportion . . .

Combined Data Analysis . . . . . . . . .-.

Intensifier Ratios . . . . . . . . . . . .
Mood Scale Ratings . . . . . . . . . . . .

Task Evaluation Ratings . . . . . . . . . .

iv

28

29

30

31

33

3Q

35

36

36

38

41

#3

4H

LIST OF APPENDICES

APPENDIX I: TIME 1 ENCODING TASK INSTRUCTIONS . . .

APPENDIX II: TIME 2 ENCODING TASK INSTRUCTIONS . .

APPENDIX III. MESSAGE CODING INSTRUCTIONS AND
CODING FORMS O O O O O O O O O C O 0

APPENDIX IV: CODING RELIABILITIES . . . . . . . . .
APPENDIX V: DEPENDENT MEASURE STANDARD DEVIATIONS .

APPENDIX VI: ANALYSIS OF VARIANCE FINDINGS . . . .

CHAPTER I
INTRODUCTION

The Problem

The area of "counter-attitudinal encoding" has stimulated a large
number of studies in recent years (Janis and Gilmore, 1965; Carlsmith, Collins
and Helmreich, 1966; Rosenberg, 1966; Linder, Cooper and Jones, 1967).
Typically, experimentation has been directed toward the assessment of atti-
tudinal effects on an individual who takes a belief-discrepant position. Sub-
jects have generally been assigned the task of encoding some "persuasive"
message which argues against an attitude which the subject either has previously
eXpressed or is assumed to hold.

The nature of the encoding task may vary widely. The required message
output may be written or oral, may allow anonymity or demand personal identi-
fication, or may involve encoding a short outline or a comparatively long essay.
The experimental situation itself may require the actual production of a
message or merely some public or private commitment to encode. In any event,
the actual communication output of such experimentation has customarily been
of secondary interest. The primary focus of counter-attitudinal encoding
research has been on the change in an individual's attitudinal position after
encoding.

One result has consistently been found: individuals will alter their

initial attitudes to conform more closely to the communication output required

of them. Advocacy of a belief-discrepant position somehow results in
attitudinal shift in the direction of that position. Much of the available
research has set the goal of determining the variables which affect the mag-
nitude of this resultant shift.

Experimentation has recently centered on the amount of attitude change
(so-called "selfepersuasion") as a function of the degree of justification
offered (differences in monetary compensation or sponsorship favorability).
The result of this inquiry has been the current controversy between pre-
dictions of shift based on dissonance theory (Festinger, 1957; Carlsmith et al.,
1966) and those based on various incentive theories (Janis and Gilmore, 1965;
Rosenberg, 1966).

The emphasis on attitudinal shift has led to an almost complete disregard
for the nature of the messages actually produced. Some attention has been paid
to a few rather gross ratings and measures of communication output. Carlsmith,
Collins and Helmreich (1966) used four ratings of essay performance: emphasis
used when making points, elaboration and originality, overall quality and per—
suasiveness, and apparent effort involved. Linder, Cooper and Jones (1967)
looked at differences in one objective measure (number of words per essay) in
addition to a number of rating scales (degree of organization, persuasiveness,
extermity of position advocated). In neither case did the authors find
differences in measures or ratings of message content as a function of justi-
fication or incentive.

Berger (1968) has provided a summarization of the message analyses which
have been performed on communication output by a number of recent researchers.
From this summary, it may be seen that there are a number of problems with
those studies which have attempted to assess the possible message consequences

of counter-attitudinal encoding.

Few objective measures have been used. Of the seven studies cited by
Berger which used some message analysis scheme, six used an Objective measure.
For four of these six, the only reported measure was essay length (number of
words per essay). The other measures noted were time spent encoding and number
of arguments used. No other objective measures of communication content, perb
formance or style were apparently used.

In addition, the ratings of message output used by such researchers as
Rosenberg (1966) and Linder, Cooper and Jones (1967) have generally been confined
to some sort of overall, composite evaluation of message "persuasiveness" or
"quality" made by two or three judges.

The researchers thus appear to have overlooked a crucial problem. Their
studies have not investigated the differences in encoding performance and
communication output resulting from an individual being required to assume a
beliefediscrepant position -- that is, the message variables which might be
affected by differences between an individual's own attitudinal position and
his assigned encoding task. i

The questions toward which the present study is directed are as follows:
How does the process of encoding a counter-attitudinal communication differ
from the process of encoding an attitudinally-consistent message? How do
people alter their communication performances and styles when writing a.belief%
discrepant message? The nature of the material encoded should be affected by
whether or not the encoder agrees with the required message output. It seems
logical that requiring a person to write counter-attitudinally should result
in some departure from his "normal" form of writing. What does an individual
do differently when he is writing something he does not agree with -- that is,
how does counter-attitudinal encoding compare with attitudinally-consistent

encoding? Some probable message and performance differences may be suggested from
an examination of related research literature.

Psychological Stress: Theory and Research

To the extent that the requirements of a counter-attitudinal encoding
situation generate some amount of psychological imbalance and stress, results
analogous to those of studies of motivation or stress effects on encoding per-
formance would be eXpected. It would appear that the inconsistency resulting
from compliance with an attitudinally discrepant task induces some ferm of
stress. Research has indicated that this heightened psychological stress should
then be reflected in an individual's performance.

Stress has customarily been conceived of as some deviation from the normal
"homeostatic" psychological condition, whether such deviation is brought about
by the introduction of psychological or cognitive inconsistency or by various
situational factors (extraneous noise, time limitations, etc.). Festinger
(1957) has stated that any departure from this normal, balanced state is somehow
psychologically uncomfortable and will result in an effort to relieve the dis-
turbance. Osgood (1957) has maintained that heightened motivation has both
generalized energizing effects and specific cue effects. Therefore, theoretically,
psychological stress should lead to certain behaviors designed to lessen this
stress.

Lazarus, Deese and Osler (1952) have summarized a number of studies
relating psychological stress to verbal and perceptual-motor behavior. In most
instances, the result of increased stress has been a decrement in performance
(in terms of slower reaction times and impairments in the quality of performance).
Thus, the general finding has been that stress has debilitating effects on
performance.

It would therefore be expected that encoding behavior should similarly

show evidence of some performance decrement under psychologically stressful

conditions. Lynch (1963) has indicated that communication output constitutes

a mechanism for tension reduction and stress relief. As such, certain character-
istics of a message should be affected by the degree of tension under which the
message is produced. If the message production task does serve to reduce tension,
this reduction of tension should then be exhibited in the behavior of the
individual encoder. Psychological stress should somehow be reflected in an
author's communication output.

A number of researchers have proposed communication output measures which
might be applied to the determination of Specific encoding style and performance
variables related to psychological stress. Gynther (1957) reported an index
termed "communicative efficiency," a measure of the relevancy of output to the

topic assigned. Dollard and Mowrer (19“?) proposed a "discomfort-relief quotient"
which was purported to index the general drive or tension state of the encoder
(with a greater proportion of "discomfort" statements indicative of higher tension).
Boder (19uo) advocated the use of an "adjective-verb quotient" as a measure of

time general activity level in a message. These and other objective measures

of the drive level present in a communicator were applied in a message analysis
conducted by Osgood and Walker (1959).

In an attempt to determine the effects of extreme levels of motivation on
encoding performance and style, Osgood and Walker examined the message
characteristics of suicide notes. The authors, utilizing some of the measures
mentioned above, compared suicide notes with ordinary personal letters. Pour
general categories of message variables were examined: stereotopy of language,
language disorganization, message indicants of the particular drive (self-
destruction) operating, and evidence of conflict. Predictions were made in

accord with the notions that psychological stress should, in general, have

debilitating effects on encoding performance and should, more specifically,
give rise to manners of expression which would reflect the psychological im-
balance state of the encoder.

Osgood and Walker found significant differences in encoding style in three
of the four general areas. Suicide notes indicated a higher level of language
stereotopy (lower type-token ratios, more repetitions, higher "cloze" scores),
thus indicating greater redundancy and predictability of message output. Suicide
notes also provided evidence of their authors' heightened drive and conflict
states (higher discomfort-relief quotients, greater qualification of verb
phrases, more ambivalence constructions and ambivalent assertions).

Although Osgood and Walker could find no evidence of structural disorgani-
zation (grammatical and Spelling errors), other reserach has provided evidence
for expecting some resultant disorganization of message output as a function

of stress.

Stress Effects on MessageProduction

Lazarus, Deese and Osler (1952) have reported studies which indicated a
deterioration in verbal production as a result of stress. Increased stress was
found to give rise to slower reaction times, longer performance times, and
increased errors in production, learning and recall. In addition, Greenberg
and Tannenbaum (1962) have investigated the effect of cognitive stress on
encoding performance. The authors found that messages produced under situations
of cognitive stress (induced by requiring a form of counterbattitudinal encoding
from subjects) took longer to encode, and contained more Spelling, punctuation

and grammatical errors.

Bettinghaus and Preston (196“) also examined the effects of psychological
stress on communicators. The authors had subjects encode single sentences in
a specified direction (either favorable or unfavorable) with respect to a number
of topics. Each subject was required to produce some statements with which he
agreed and some with which he disagreed. The authors reported an overall
tendency for individuals operating under the pressures of cognitive inconsistency
to take more time to encode than those same individuals under cognitively con-
sistent conditions.

Thus, in summary, messages produced by people under conditions of heightened
motivation or tension (due to the introduction of variables tending to increase
psychological imbalance) should exhibit certain measurable tendencies which are
indicative of an overall performance decrement. The encoding performance itself
should take longer (or proceed at a slower rate) and the message output should

contain more errors.

Language Intensity: Related Research

Messages encoded under conditions of psychological stress should also
exhibit characteristics which may be said to indicate decreased language intensity.
It has been found that prior attitudes influence recall of information related
to those attitudes (Klapper, 1960) as well as the subsequent transmission of
this information (Allport and Postman, 19u5). It would therefore be expected
that prior attitudes should influence any behavior relevant to these attitudes.
The process of encoding a message in opposition to an individual's prior
attitudinal position should differ from that of encoding a message which is
consistent with one's prior beliefs. These differences should be manifested in a

greater hesitancy and a lesser intensity in the assertions made by the encoder.

It is posited that an encoder who disagrees with what he is saying would be
less likely to use emphatic, extreme language than would an encoder who was
in full agreement with the views he was instructed to write on.

Thus, in addition to the performance characteristics expected to result
from the heightened tension accompanying a psychologically stressful encoding
situation, the style of the message (in terms of expressed language intensity)
should also be reflective of an inconsistency between prior attitude and required
output...

A number of indices of what may be termed language intensity can be suggested.
The level of language intensity in an assertion is partially indicated by the
extremity or polarity of the modifiers in that assertion. Howe (1962; 1966),
Lilly (1968a; 1968b) and Cliff (1959) have scaled a number of adverbial
modifiers in terms of the degree to which they modify the perceived extremity
of various adjectives, as judged by a group of receivers. Their results indicated

that varying degrees of polarity are expressed by modifiers of the following

 

 

 

types:

Modifier Type Low Intensity high Intensity

Probabilistic doubtfully certainly
conceivably undeniably
perhaps definitely

Intensive slightly extremely
somewhat very

Frequency rarely repeatedly
occasionally frequently
sometimes usually

Temporal Proximity eventually instantly

subsequent 1y immediately

Osgood (1959) has proposed a measure of assertion intensity labelled
"evaluative assertion analysis." As a method of indexing the degree to which
the author of a given message deviates from evaluative neutrality in discussing
a concept, the method relies, in part, on the extremity of the modifiers
(adverbs and adjectives) selected by the author. Osgood has reported that
judges can reliably distinguish between modifiers indicative of polarized or
extreme attitudes (:_3 modifiers such as "absolutely" or "extremely") and those
which suggest a less extreme position (:_1 modifiers such as "partially" or
"somewhat"). Therefore, one index of the degree of favorabllity or unfavorability
expressed in a message would be the polarity of the modifiers chosen by the
source in assertions related to the message topic.

Another message characteristic presumably reflective of the attitudinal
position of the source would be the yerb§_used in the assertions made by that
source. Osgood stated that "connectors which imply only possible or hypo-
thetical relations between actor and complement" are expressive of low (I l)
intensity (e.g. "may," "might"). Moderate (1_2) intensity is connoted by
most modal auxiliary verb ferms (e.g. "used to," "would," "plan to"). Verb
forms indicative of high (1_3) intensity were said to be those in the un-
qualified, simple present tense (e.g. "is," "must," "demands").

Another indicator of the source's degree of favorability or unfavorability
toward a topic should therefore be the polarity of the verbs contained in the
source's communication output. This polarity appears indicable by the general
mood of the verbs employed. High intensity language would be marked by less
use of conditional, subjunctive verb forms (e.g. "may ," "might," "could,"

"seems") and a corre5pondingly greater use of the more polarized simple indicative

10

or imperative mood (e.g. "must ," "have to," "is ," "demands ," "has").

The tense of the verb has also been shown to influence the level of
perceived favorability in an evaluative assertion. Osgood has reported that the
present tense is indicative of higher intensity than the past or future tenses.
Holsti (1967) stated thr:the use of the past tense is expressive of a lower
level of evaluative intensity than is the use of the present tense. Howe (1966)
demonstrated that past and future tenses both connote decreased intensity of
connection from that manifested by the simple present tense.

The voice of the verbs in a given message has also been found to influence
judgments of the activity level of the language used. Johnson (1967), drawing
upon some earlier work by Clark (1965), investigated the differences in ratings
of terms in active and passive sentences. He found that the use of verbs
in the passive voice altered subsequent ratings of the object of the sentence
from that obtained when active voice verbs were med. Specifically, the concept
being considered by the source was judged as being less dynamic (softer, weaker,
less aggressive, less energetic, less potent) under conditions of passive voice
verb constructions.

In summary, it may be said that variations in the apparent mood, voice
and tense of the verb are all indicators of some deviation from evaluative
neutrality in a source's treatment of a message concept.

Osgood and Walker (1959) have reported a greater percentage of "ambivalence
czonstructions" used by encoders under psychological stress. Several of these
sample ambivalence constructions ("maybe ," "possibly ," "seems ," "appears")
correspond to what has been discussed above as indicants of low language intensity.
Others of these constructions ("but," "however") may be viewed in terms of the

discussion of message indexing cues put forth by Tannenbaum (1955). Tannenbaum

11

found that the insertion of the conjunction "but" in a message increased the
probability that the message would be viewed as a two-sided presentation. It
would seem that the apparent two-sidedness of a communication would serve as
an additional indicant of the degree of opinion polarization in a source.
Greenberg and Tannenbaum (1962) reported an increase in the likelihood of
encoding two-sided messages when performing under conditions of counter-
attitudinally induced stress. Thus, some index of the amount of polarity,
reflected in the use of ambivalence constructions, would appear to result
from psychologically stressful antecedents to message production.

McEwen and Greenberg (1969a; 1969b) reported that messages containing low
intensity modifiers and verb forms were judged to be lower in intensity.

These messages were rated as expressive of greater hesitancy, uncertainty and
weakness of tone than the same message contan:containing high intensity

modifiers and simple, present tense verb forms. Thus, it appears that the

. modifiers, verb forms and ambivalence constructions present in a message indicate
the apparent intensity level of the attitudinal position of the message source.
Since the hesitancy expressed in a message should somehow be reflective of the
hesitancy of its author, these language style variables could be used to assess
the level of uncertainty and intensity of feeling present in the writer.

It fellows that sources required to encode counterbattitudinal arguments
should be experiencing some hesitancy as well as some psychological stress or
tension. It is the contention of the present study that the communication out-
put of an encoder will be affected by the consistency or lack of consistency

between the requirements of the task and the encoder's own attitudinal position.

12

Readabilij' Research

Gynther (1957) reported that subjects in a high anxiety situation produce
more disorganized, disrupted (less "efficient") communication output. This is
consistent with Clevenger's (1959) summary of SXperimental research on tension
or stage fright effects on Speakers. Clevenger cites research literature which
has atteupted to quantify the disruptive effects of tension on communicators.

It seems that disorganized communication (e.g. run-on sentences) would be more
difficult to understand or comprehend. To this extent, a parallel can be found
in the various discussions of what has been termed readability (Flesch, 19u9;
Klare, l963). Readability has been proposed as an index of the ease with which
potential readers can follow or understand a given message. Readability formulas
(e.g. the FleSch measure) basically rely on two measures: average sentence
length and number of syllables per hundred words.

In applying the prediction that cognitive stress leads to disruption of
message output, Greenberg and Tannenbaum (1962) found that counter-attitudinally
induced stress led to lower message readability scores. Following exposure to
a cognitive stress-inducing situation, individuals encoded messages which con-
tained more words per sentence and more syllables per hundred words.

Therefore, an additional message style measure which should be affected
by the correspondence (or lack of correspondence) between required output and

prior attitude would be the readability level of the message.

Hypotheses

 

There would appear to be at least two general dimensions along which
encoding differences would be expected as a result of inconsistencies between

source attitude and required message output: message production variables

13

(variables indicative of the rate and quality of message output) and message

style variables (those indicative of differential style, expression or manner

of content presentation).

Message Production Variables. It is predicted that the introduction of

psychological stress or tension (brought about by the counter-attitudinal

encoding requirements) will result in a general impairment of encoding per-

formance.

Hypothesis 1:

Hypothesis 2:

Message Style

Counter-attitudinal encoding will result
in Slower rates of message production than
will attitudinally-congruent encoding.

Counter-attitudinal encoding will result

in higher frequencies of message errors and
alterations than will attitudinally-congruent
encoding.

Variables. It is predicted that the writing style of

 

encoders performing

a counter-attitudinal task will differ from that of encoders

writing under cognitively conSiStent conditions.

Hypothesis 3:

Hypothesis u:

Counter-attitudinal encoding will result
in a decrease in the level of language
intensity from that obtained under
attitudinally-congruent conditions.

Counter-attitudinal encoding will result
in lower readability scores than will
attitudinally-congruent encoding.

CHAPTER II

METHODS AND PROCEDURES

Overview

All experimental subjects wrote two messages, one which argued in
accordance with and one in opposition to a previously stated opinion. En-
coding style and performance variables were compared for each subject arguing
counter-attitudinally and attitudinally-consistent, while controlling for order
of encoding and attitude toward the issue. Each subject's mood and evaluation
of the task was assessed immediately after completion of the required message

tasks.

Subjects

Subjects (N=76) were obtained from an introductory business writing class
at Michigan State University. The experimental encoding requirements were
Similar to the prior classroom experiences of these students, thus helping

to provide motivation for satisfactory task completion.

Design and Analysis

 

Two independent variables were used: encoding conditions and topic
evaluation.. Each subject wrote messages in each of the two encoding conditions
(counter-attitudinal and attitude-consistent). In addition, half the subjects
addressed a positively evaluated topic, while the other half encoded messages

concerning a negatively evaluated topic.

1n

15

The overall design is in Figure l, with each subject producing two

messages, one counter-attitudinal and one attitude-congruent.

 

Encoding Conditions

 

 

 

 

Counter- Attitude-
Attitudinal Consistent
Positive
Topic (Group I)
Evaluation
Negative
(GrouP II)
Figure 1.

 

.Message style and performance measures were obtained for each message in
each of the four cells. Comparisons in terms of the dependent measures were

made between the two encoding conditions. Results were analyzed separately for

the two tapics.

Procedures.

 

The experiment was preceded by a pretest which determined two suitable
issues for the message encoding task.

Pretest Procedures. Two weeks befOre the actual experiment, a different
group of subjects (N=7u) completed a topic pretest. The pretest consisted of
Six Opinion statements accompanied by five-interval rating scales (ranging from
"strongly agree" to "strongly disagree"). Subjects indicated the extent of

their agreement or disagreement with each statement. From this pretest, two

16

issues were selected for inclusion in the experiment. The topics chosen were
the twO most polarized (one most negatively, the other most positively). The
most highly favored issue was: "The federal voting age should be changed so

as to permit those 18 years of age and older the right to vote." The mean
rating of this topic was 3.59 on afive-point scale, with a rating of one
indicating strong disagreement and a rating of five indicating strong agreement.
The most highly opposed issue was: "Television programming Should be censored
by a federal control board." This topic had a mean rating of 2.01 on the same
five-point scale.

Experimental Procedures. Subjects (N=76) were randomly assigned to one

 

of two experimental groups. Group I subjects (n=l+2) were instructed to encode
messages (one message in opposition to the topic and one in support of the
topic) which dealt with the "18 year old vote" issue. This was the issue
toward which most subjects had expressed a favorable opinion. Group II subjects
(n=3it) encoded messages arguing on both sides of the "federal TV censorship"
issue, the one toward which subjects had expressed a generally unfavorable
attitude.

‘ The order of encoding was counter-balanced in each group. Half the
subjects in each topic group encoded counter-attitudinally at time 1 and
attitudinally-consistent at time 2. The other half of each eXperimental group
followed the opposite order.

Before the first encoding task, subjects reSponded to the two concepts
selected from the pretest on five evaluative semantic differential scales:

good-bad, valuable-worthless, fair-unfair, wise-foolish, and important-unimportant.

Only one concept was relevant to a given subject's experimental treatment.

17

Next, subjects encoded the first message required by their topic-condition
and order sub-group. Each subject was instructed to write a highly persuasive
message which would be capable of convincing potential readers that the subject's
assigned position was the more reasonable and justifiable one. The encoding
task was introduced as part of a Study on the way people use arguments in
attempting to create effective messages of opinion, such as might be found in
newspaper editorials. Subjects were given no time limit. The instructions
used are in Appendix 1.

Subjects then were asked to respond to a number of mood scales (Greenberg,
1960) consisting of Single adjectives accompanied by five-interval rating scales,
ranging from "definitely applies to my mood" to "does not apply." Sample mood
scale adjectives employed were: "insecure ," "angry," "startled," and "tired."
Subjects were also asked to rate the task on two five-interval rating scales
("difficult-easy" and "pleasant-unpleasant").

In the next class period, subjects encoded a message arguing on the opposite
side of the issue which they had previously written about. Again, subjects
were asked to write as persuasive a message as they felt they could and were
permitted sufficient time to complete the task to their satisfaction.

Subjects then responded to the same mood and task evaluation scales which

had been given at time 1. These instructions are in Appendix 2.

Independent Variables

Two independent variables were employed in the experiment. These were the
conditions (that is, counter—attitudinal or attitude-consistent) in which subjects
did their encoding and their evaluation of the topic to be written about

(that is, whether subjects had a positive or a negative attitude toward the encoded

issue).

18

Encoding Conditions. Each subject wrote both in congruence with and in

 

opposition to his prior beliefS. The experimental measure of attitude toward
the topic provided a means of checking the extent of each subject's evaluative
polarity with respect to the issue he was writing about. This measure con-
sisted of the two selected issues ("18 year old vote" and "federal TV censor-
ship"), each accompanied by the five 7-interval rating scales mentioned above.
Individual encoding characteristics thus were compared for each subject encoding
under the two required encoding conditions.

Evaluation of the Topic. Group I subjects encoded messages arguing on

 

both Sides of an issue which was highly favored by these subjects. Group II
subjects encoded messages which addressed an issue toward which the majority
of subjects had expressed strong opposition. In this manner, individual en-
coding Style and perfOrmance variables were then compared for each subject
arguing counter-attitudinally as opposed to attitudinally-consistent, while
controlling for the direction (positive or negative) of this argumentation.
For half the subjects (Group I) counter-attitudinal encoding consisted of
writing a proposition-attacking message. For the other half (Group II), the
counter-attitudinal task required was that of writing a proposition-supporting

message.

Control Variable
One variable was controlled -- the order in which the messages were encoded.
Order of Encoding. The possibility of encoding order effects was con-
trolled for by randomly assigning subjects in each experimental group to two
alternative encoding-order procedures. Approximately half the subjects in each
group wrote counter-attitudinally at time 1 and attitudinally-consistent at

time 2. The remaining subjects encoded in the reverse order.

l9

DEPENDENT VARIABLES

Pe rforman ce Vari ab leS

 

Two major measures of encoding performance were used. The rate of
encoding (words per minute) and the number of encoding errors were ascertained
for each message produced by each experimental subject.

Encoding Rate. Subjects were timed.while performing the encoding task.
After having filled out the experimental pretest, subjects all began the
encoding task at the same time. Upon completing the message task, subjects were
requested to raise their hands. At this time, the mood and task evaluation scales
were given to each subject. Experimenters noted the time (in minutes) taken by
each subject in writing his assigned message.

The length of the message (in total words) produced by each subject was
ascertained and the individual encoding rate (in words per minute) was computed.

Encoding Errors and Changes. Three indices of the errors or changes made
by subjects in the process of message production were employed. The number of
Spelling errors and the number of punctuation errors in each message-was
determined. In addition, the amount of editing apparent in each message was
ascertained.

a. spelling errors. The number of spelling errors made by each subject
in each message was determined. A spelling error was here defined as any
Spelling of a word other than that commonly accepted in a standard English
dictionary.

b. punctuation errors. The number of gross punctuation errors made by
each subject was determined fer each message. Only major errors, such as
failure to use periods or question marks when required, failure to capitalize

proPerly, and omission of apparent commas and colons were counted.

20

c. editing. The number of occasions of editing within each experimental
message was determined. Counted as indices of editing were: cross-outs of words
or parts of words, evident corrections or re-tracings of letters or entire words,
obvious erasures of words or parts of words, insertions of words or parts of

words.

Message Style Variables

Two major variables were used as indices of the language style used by
each encoder. These were the readability of each message and the level of
language or message intensity.

Readability. The two indices of a message's level of readability originally
suggested by Flesch (19u9) are the number of words per sentence and the number
of syllables per hundred words. For each experimental message, both indices
were computed. The average number of words per sentence was determined and the
number of syllables in the first hundred words of each message was ascertained.

Language Intensity. The polarization of the modifiers, the use of verb
forms and the presence of ambivalence constructions in the messages written by
the experimental subjects were measures of the language intensity level of
each message.

a. modifier polarization. The number of high, moderate and low intensity

 

qualifiers in each message was determined and the proportion of each modifier
'type (per total words) was computed.

The following general criteria were used in classifying adverbial or
adjectival modifiers. High intensity modifiers were defined as those indicating
extreme intensity, probability, frequency or temporal proximity. These were
classifiable on the basis of quantitative values reported by Cliff (1959),

Howe (1962; l966), Lilly (1968a; 1968b) and Dodd and Gerbrick (1960). Ex-

21

amples of modifiers coded as high in intensity were: "extremely," "definitely,"
"certainly," "immediately ," and "continually"). In addition, superlative forms
of adjectives (e.g. "greatest," "biggest") were considered as highly polarized
modifiers.

Moderate modifier polarization was reflected in the use of indefinite
adverb and adjective qualifiers (such as "much" or "many"), by moderate level
probabilistic, intensive, frequency or temporal proximity modifiers (e.g. ,
"rather," "frequently," "soon"), and by comparative forms of adverbs and
adjectives ("bigger," "happier").

Qualifiers of low polarization were defined as those indicating very low
intensity, probability, frequency or temporal proximity (such as "slightly,"-
"partially," "sometimes ," "conceivably," "eventually").

b. use of verb forms. Three linguistic criteria were used to indicate

 

the level of intensity in the verb forms of a given message. The apparent
mood (held to be indicative of the polarity of the verb form), voice and tense
of each verb were used to represent some aspect of the degree of definiteness
and language intensity in the subject's communication output.

(a) verb polarization. The proportions of high, moderate and low
polarity verbs (per total words) were computed for each message. Verbs
implying conditionality (possibility, potentiality) were considered as ex-
pressive of low polarization or intensity. Verbs in the subjunctive mood,
compound verbs with the modal auxiliary form indicative of hesitancy or con-
ditionality (e.g. "might ," "seems to," "appears to") and verbs in clauses pre-
ceded by "if" were classified as low in polarity. Clauses preceded by "perhaps"

or "maybe" had the intensity values of the verbs contained in those clauses

22

reduced one level. In such instances, high intensity verbs were reduced to a
rating of moderate, and moderate or low intensity verbs were reclassified as
being of low polarity.

Moderate levels of verb polarity were indicated by qualified indicative
mood forms wherein the auxiliary verb was not classifiable as low or high in
apparent intensity. Examples of moderate intensity verbs are "ought to,"
"would" and "should" (that is, verb forms expressive of a higher level of
probability and potentiality than that reflected by low intensity verbs),

High polarity verbs were those in the unqualified indicative mood (e.g.
"to be ," "to have," "to oppose," "to go"), verbs in the imperative mood
("bewardl"), or those accompanied by auxiliary verbs which imply obligation
or necessity (e.g. "must ," "demands," "requires"). In addition, negation of
several conditional verbs ("can not," "would not") resulted in the verbs being
categorized as expressive of high intensity.

Complete coding instructions for the classification of the polarity
level expressed by both modifiers and verbs may be found in Appendix 3.

(b) gigs. The number of passive voice and active voice verbs in

each message was determined and an activity ratio (Na/Nam where Na was

p’
the total number of active verb forms and Np was the total number of passive
verb forms in each message) was computed.

(c) 13223.9: The numbers of past, present and future tense verb forms
present in each message was determined and the ratio of each verb tense (to
total verbs) was calculated.

c. ambivalence constructions. The final index of the language intensity

level in each message was the proportion of ambivalence constructions

23

(conjunctions of the form: "but ," "Still," "however," "although") per total

words .

Individual Task and Mood Scale Ratiggg

Experimental subjects rated their personal feelings and reactions to
their assigned tasks immediately upon completion of each encoding assignment.
These measures served as a partial check on the adequacy of the experimental
manipulations in inducing stress or anxiety.

Mood Scale Rating_. Subjects were asked to complete fifteen rating
scales designed to index their mood or feelings when in the act of encoding.
These rating scales consisted of Single adjectives accompanied by five-interval
rating scales on which subjects indicated the extent to which each adjective
applied to their personal mood at the time of encoding (from "definitely
applies to my mood" to "does not apply"). Mood adjectives were those used
by Greenberg (1960) in assessing the extent of stress among groups of encoders.
The author reported that five factors seemed most successful as indicants of
cognitive stress. These factors, and the adjectives which best represented

them, were:

Aggression angry, defiant, rebellious
Anxiety: startled, fearful, insecure
Deactivation: drowsy, sluggish, tired

Social Affection: affectionate, forgiving, kindly
Pleasantness: pleased, lighthearted, refreshed

Increased stress is presumed indicable by increases in subjects' scores
on the aggression and anxiety factors, and decreases on the deactivation,

pleasantness and social affection factors.

2')

Task Evaluation Ratings. After encoding each message, subjects rated

 

the communication task on two 5-point rating scales: "very pleasant-very

unpleasant" and "very difficult-very easy."

Readability,

 

A coding form Specifying the rules for classification of language intensity
(both in verbs and modifiers) as well as the rules for syllable counts, errors
and so forth was constructed, based on the general descriptions provided above.
The form may be found in Appendix 3. Inter-coder reliabilities were determined
by correlating alternative codings of randomly selected messages. This was

done as a partial check on the coding form and the rules for classification.

CHAPTER III

RESULTS

"7

This study examined the effects of counter-attitudinally induced stress
on an individual's communication output. It was postulated that an increase
in psychological stress would be manifested in specific changes in communi-

cation performance and style variables. In comparison with an individual's

 

attitude-consistent performance, it was hypothesized that counter-attitudinally
encoded messages would be produced at Slower rates and would contain more en-
coding errors. It was further predicted that individuals, when writing in
opposition to their prior attitudes, would produce messages which were less
readable and contained elements indicative of lower levels of language inten-
sity than messages produced by these same individuals under attitudinally-
consistent conditions .

Each subject wrote two messages. One message argued in accord with a
previously stated attitude. The other message argued in opposition to this
attitude. Comparisons were made between the individual's performance and
Style characteristics under attitudinally-consistent conditions and those

measures obtained when the individuals encoded counter-attitudinally.

Subject Attrition

 

Subjects (n=20) who wrote only one experimental message (that is, those
who did not Show up for part two of the experiment) were eliminated from the

25

26

analysis. The 20 were distributed equally from the two encoding topic
conditions.

In addition, subjects (n=ll) who, on the basis of the experimental pre-
test topic rating, had a neutral attitude (mean rating of u.o over the five
7-point scales) toward their assigned topic were eliminated from further
analysis. There did not seem to be any basis for determining which messages
were counter-attitudinal and which attitude-consistent for these subjects.

Therefore, in the analyses, one group (the "18 year old vote" group)
contained 26 subjects and the other group (the "federal TV censorship" group)

had 19 subjects.

m Reliabilities
Forty messages of the 90 total, were randomly selected for either complete
or partial recoding in order to assess intercoder reliabilities. The author
originally coded all messages, while another coder was trained and did the
recoding. At the time of coding, messages were not identified as to whether
they had been produced by subjects arguing counter-attitudinally or attitudinally-
consistent, so neither coder was aware of the experimental condition for a
given message. Also, the second coder was unaware of the experimental hypotheses.
The results indicated a reliable coding procedure had been developed.
Correlations of alternative codings of each of the nineteen dependent measures
, showed relatively high inter-coder agreement in almost every case. With
reSpect to the coding of verb and modifier polarity (held to be critical as
perhaps having a greater Opportunity for subjective rating), the lowest of the
twelve obtained correlations was .815 and eight of the correlations were above

.90.

27

Only twice (of 33 checks made) did any obtained correlation fall below
.80 (.56“ for numbers of punctuation errors and .605 for numbers of passive
voice verbs). In each case, the instances of coder disagreement and coding
form ambiguity were discussed and rectified. Repeat correlations on additional
sample messages provided acceptably high correlations (.88“ for punctuation
errors and .891 for passive voice verbs). Complete listings of the obtained

reliability results are found in Appendix 1:.

Performance Variables

The dependent measures of performance employed were encoding rate and
message errors.

Encoding Rate. No Specific time limit was imposed on the subjects. Of
the total time available in the 50-minute class periods, the maximum time used
by any one subject was 38 minutes. This would indicate that little if any
time pressure seems to have Operated on the encoders.

The mean rates of message production, in words per minute, are in Table 1.
There was no significant difference in the rate at which individuals produced
messages as a function of the consistency of required encoding performance with
prior attitude. Also included in Table l are the mean times to encode, Since
subjects arguing counter-attitudinally could have produced messages at the same
rate, but for less overall time than when they were producing attitude-con-
sistent messages. No Significant differences in total time spent encoding were

observed, however.

28

 

 

 

 

Table 1.
Required Message Output
Message
Topic Counter-Attitudinal Attitude-Consistent
"18 year 12.“ ll.9
old vote"
Mean
Encoding Rate
"federal TV 11.8 t 11.6
censorship
Total
Time to Encode
"18 year 22.2 22.5
old vote"
"federal TV 21.7 22.u
censorship"

*Standard deviations for all variables are in Appendix 5.

 

Message Errors. Three main types of message errors and changes were
used as indices of the possible effects of counter-attitudinally induced
stress on encoding: punctuation errors, spelling errors and editing (cross-
outs, erasures, insertions). In addition, an overall error score for each
subject was computed by summing across the individual error indices. The re-
sults of a correlated t_comparison of subject scores for the four error and
change measures appear in Table 2. Overall, in seven of the eight comparisons,
the direction of the difference is as predicted, although none of the differences

achieve the .05 level of Significance.

29

 

Table 2. Message Errors

 

Required Message Output

 

 

 

 

Message
Topic Counter-Attitudinal Attitude-Consistent
Punctuation "18 year .017 .01u*
Errors old vote"
per 4
Total Words "federal TV .022 .021
censorship"
Spelling "18 year
Errors old vote" .009 ,oo7**
per
Total Words "federal TV .016 .015
censorship"
Editing "18 year .ou7 .050
per old vote"
Total Words
"federal TV .101 .063*
censorship"
Total "18 year .o7u .072
Errors per old vote"
Total Words
"federal TV .139 .100*
censorship"

 

* p (.15, by correlated t-test (one-tailed probability)

** p<.10

 

Punctuation Errors.

The difference in individual counter-attitudinal

and attitude-consistent message performance with respect to the number of

punctuation errors committed per total words was not Significant, although in the

predicted direction (p ( .15).

3O

Spelling Errors. The counter-attitudinal and attitude-consistent messages
did not differ Significantly in the proportions of Spelling errors, although
the differences were in the predicted direction. In the case of the proposal
of the 18 year old voting age, the probability value of the obtained correlated
3 approached Significance (p_(.10).

Editing. No statistical differences in the amount of total editing changes
were observed.

Total Errors and Chggss. A tendency (p( .15) was again observed for
individuals (in both topic conditions) to make more total message errors when
arguing counter-attitudinally than when writing messages with which they agreed.

Errors_per Minute. In order to determine whether subjects possibly made
more errors per unit of time Spent encoding, the total number of message errors
and changes made by each subject was divided by the amount of time the subject
Spent encoding. The results are in Table 3. In both topic conditions a ten-
dency to make more errors per unit of time Spent encoding (py(.lS) was noted
when subjects were encoding attitude-discrepant messages. Thus, there seems
to be some evidence of performance decrement in terms of error rate (errors
per minute) which occurs when individuals' counter—attitudinal.writing perb

formances are compared with their attitude-consistent message performances.

 

Table 3. Message Error Rate

 

Required Message Output

 

¥__: Counter-Attitudinal Attitude-Consistent
"18 year
old vote" .876 .80u*
Message
Topic
"federal TV 7 1.1u1 l.003*

censorShip"
I p.(.15, by correlated t-test

31

Mess age Sgle Variables
Comparisons of the two major categories of Style variables (readability
and message intensity) are presented below.

Readabili_ty. No measures of readability were used in the present

 

study: mean sentence length and total syllables per first 100 words. Results

of the readability comparisons are in Table u.

 

Table 1:. Readability Measures

 

 

 

Message Required Message Output
Topic Counter-Attitudinal Attitude-Consistent
Mean "18 year 20.8 21.u
Sentence old vote"
Length
"federal TV 18.3 18.3
censorship"
Syllables "18 year _, 149.6 ' 147.6
per 100 old vote" '
Words
"federal TV 158.2 155.2
censorship"

 

a. sentence lengfl. No Significant differences in the numbers of words
per sentence were found between the message conditions.

b. syllables per first hundred words. No Significant differences were
noted in the numbers of syllables in the first hundred words of the experimental
messages.

Language Intensifl. Three chief indices of the level of intensity used
by each encoder in writing his two required messages were used: modifier

polarization (prOportions of high, moderate and low polarity modifiers), use of

32

verb forms (polarization of encoder language indicable by verb polarity, voice
and tense) and ambivalence constructions.

a. modifier polarization. Results obtained from the comparisons of

 

individuals counter-attitudinal and attitude-consistent message output are in
Table 5. Contrary to hypothesis, no increased usage of highly polar modifiers 7......
was found when the individual subjects wrote attitudinally-congruent messages.
An inconsistent pattern was found in the case of low polarity modifiers.

Subjects writing on the federal TV censorship issue tended to use more low

 

intensity modifiers when writing in accord with their previously expressed L
attitudes (p(.15), whereas it had been predicted that these subjects would

use .1_e_§_s_ low intensity modifiers when encoding attitude-consistent messages.

Subjects writing on the 18 year old vote proposition, however, conformed more

with the experimental hypotheses, with a. tendency (p< .15) to encode more low

intensity modifiers when writing counter-attitudinal messages.

b. use of verb forms. Three indices of the language intensity present
in the verb forms used by each subject are presented below: verb polarization,
verb voice and verb tense.

(a) verb polarization. Comparisons were made of the extent of each
subject's usage of high, moderate and low polarity verb forms, as indexed
chiefly by the apparent mood of the verbs contained in the messages produced
by the subjects. The findings are in Table 6. As hypothesized, there was
a tendency in each topic condition for greater usage of highly polar verb
forms by individuals encoding under attitudinally congruent conditions than was
observed under counter-attitudinal conditions. In the case of the TV censor-
ship issue, this difference was p_(.15. No Significant differences were found

in the usage of low polarity verb forms.

s".

33

 

Table 5. Modifier Polarity

 

 

 

 

 

 

 

 

Proportibn
(per total words)
Message Required Message Output
Topic Counter-Attitudinal Attitude-Consistent Ian
E
"13 year .023 .022
High old vote" f
Polarity i
Modifiers "federal TV .026 .029 j
censorship" f
"18 year 2
Moderate old vote" .038 .040 3
Polarity
Modifiers "federal TV .032 .026
censorship"
"18 year .009 .007*
old vote"
Low
Polarity "federal TV .008 .010*
Modifiers censorship"
*p (.15, by correlated t-test
No prediction had been made as to the possible use of moderately polar
verb forms, but, when addressing the TV censorship topic, subjects used
significantly more (p<.05, two-tailed) moderately polar verbs when arguing
counterbattitudinally than they did when arguing attitudinally-consistent.
r

34

 

Table 6. verb Form Polarization

 

Proportion (per total words)

 

 

 

 

Message Required Message Output
Topic Counter-Attitudinal Attitude-Consistent
"18 year .067 .068
High old vote"
Polarity
Verbs "féderal TV .063 .068*
censorship"
"18 year .025 .026
Moderate old vote"
Polarity
verbs "federal TV .028 .022**
censorship
"18 year .0114 .012
old vote"
Low
Polarity "federal TV .018 .018
Verbs censorship"

 

* p (.15, by correlated t—test
** p (.05

 

(b) verb voice. Results of the verb voice comparisons are listed
in Table 7. A generally greater proportion of active verbs was used in
messages encoded by subjects writing in accord with their beliefs. This trend
was found for both the TV censorship issue (p_(310) and for the 18 year old

vote issue (p_ <. 20).

35

 

Table 7. Verb Voice

 

Proportion
(per total verbs)

 

Message Required Message Output
Topic Counter-Attitudinal Attitude-Consistent
"18 year .88 .89
old vote"
"federal TV .87 .91*
censorship"

 

*p (.10, by correlated t-test

 

(c) verb tense. An inconsistent pattern was found in terms of the

verb tense variable, as in Table 8. For subjects addressing the TV censorship
proposition, there was some tendency to use relatively more present tense verb
forms when arguing in accord with prior attitudes (p_(.20). This trend is as
predicted. For the 18 year old vote issue, however, the Opposite tendency was
found, with messages encoded under attitudinally-consistent conditions containing
somewhat E333 (p_ (.15) present tense verb forms. This latter trend is opposite
to the experimental predictions.

c. ambivalence constructions. Contrary to expectation, there was
apparently no greater usage of conjunctions classifiable as "ambivalence
constructions" when individuals argued counter-attitudinally as opposed to

attitudinally—consistent. The findings of this comparison are in Table 9.

 

36

 

Table 8. Verb Tense

 

 

 

 

 

Proportion
(per total verbs)
Message Required Message Output
Topic Counter-Attitudinal Attitude-Consistent
"18 year .850 .811*
Present
Tense
Verbs "federal TV .870 .898
censorship"
Past "18 year .120 .161*
Tense old vote"
Verbs
"federal TV .068 .058
censorship"
"18 year .029 .032
Future old vote"
Tense
Verbs "federal TV .060 .044
censorship"

 

*p (.15, by correlated t-test

 

 

Table 9. Ambivalence Construction

 

Proportion (per total words)

 

Message

Topic Counter-Attitudinal Attitude-Consistent
"18 year .006 .007

old vote"
"federal TV .003 .003

censorship"

37

Summggy

An examination of the encoding perfOrmance and style variables employed
indicates an overall tendency for the experimental predictions to be borne out,
at least in terms of direction, if not amount, of change. Across the total
of 34 possible message comparisons, the differences between the attitude-
consistent and counter-attitudinal measures were in the predicted directions

in 23 of the cases.

Combined Analysis

 

A number of the predicted relationships approached significance. Failure
to obtain statistical levels of significance could be due in part to the
relatively small number of subjects in each of the topic categories. Combining
subjects across these categories could both increase cell size and permit an
extended examination of relationships. We subsequently compared the attitude-
consistent and counter-attitudinal message outputs of all subjects, regardless
of topic category, for the following major perfOrmance and style variables:
production rate, total message errors, error rate, high and low polarity
modifiers, high and low polarity verbs, active voice verbs, present tense
verbs, words per sentence and syllables per first hundred words. The results

of this analysis are in Table 10.

38

 

Table 10. Combined Data Analysis

 

RequiredMessage'Output

 

 

 

Variable '7 '" ”‘Counter-Attitudinal Attitude-Consistent

Production Rate 12.2 11.7

(words per minute)

Total Errors .102 .084e

(per total words)

Error Rate .988 .888***

(per minute)

High Polarity .024 .025
Modifiers

Low Polarity .010 .010
Modifiers

High Polarity .065 .068
verbs

Low Polarity .015 .014
Verbs

Active Voice .872 .899**
Verbs

Present Tense .859 .848
Verbs

Words per 19.7 20.1
Sentence

Syllables per 153.2 150.8
100 Words

* p < . 15

** p <. 10

*** P (.05

 

if!
’

39

Eight of the eleven hypothesized differences are in the predicted
directions. Of the seven differences with probabilities of less than .25, Six
were in the direction hypothesized. One difference reached significance.
Subjects made significantly more (p_(.05) errors per minute of time spent
encoding when they wrote counter-attitudinally than they did when they wrote
attitude-consistent messages. Two of the differences approached Significance.
Subjects tended to make more errors per total number of words encoded (p<.15)

when writing attitude-discrepant messages. There was also a tendency for

 

subjects to use relatively more (p_(.10) active voice verbs (and, hence, fewer
passive verbs) when writing messages with which they agreed.

Thus, although the general production rate does not Show the performance
decrement usually accompanying increased stress, both total error measures
(errors per total words and errors per unit of time Spent encoding) provide
evidence that some disturbance of encoding performance has occurred in the
situation where subjects are encoding attitude-discrepant messages.

Also, there is some support for the proposition that, when arguing
counter-attitudinally, subjects use somewhat less intense language than when
they are arguing in accord with their prior beliefs. The tendency toward less
intense language is evidenced by a decrease in the percentage of active voice
verbs used (p<.10). These tendencies, while they do not reach statistical
significance at the .05 level, do Show some overall trend in the predicted

direction .

Sign—Test Analysis
There is the possibility that a small number of individuals undergoing

relatively large changes could have distorted the results so as to either inflate

40

or deflate the mean group differences. In order to determine the degree to
which the changes in each individual's communication behavior corresponded
to the experimental predictions, a Sign test (Siegel, 1956) was performed
on the data. The results of this additional analysis closely parallel
those previously reported. Combining across topic-conditions, four of ..a%
nineteen comparisons by Sign test were significant (2‘ .05) and correctly
predicted: low intensity verbs per total words, high intensity verbs per
total words, active voice verbs per total verbs, and total errors made per

minute. Therefore, all variables found to be significant in the parametric

 

analysis of combined topics were Significant by Sign test. The variable of
low intensity verbs per total words was significant in this analysis, but

not in any of the parametric analyses.

Intensifier Ratios

Our final analysis compared the percentages in each message of total
classifiable modifiers or verbs which were of high or low intensity. Rather
than determine the proportion of a given modifier or verb type per total words,
it was decided to examine the percentage of verbs or modifiers used which were
of a given level of polarization or intensity. In other words, the analysis
was designed to ascertain, for example, how often, when the subject used a
modifier it was a high intensity modifier or, when the encoder chose a verb
it was a highly polar verb. Comparisons of these "intensifier ratios"
(proportions of total verb or modifier intensifiers which were of a specific
polarity level) were made for each subject's counter-attitudinal and attitude-

cOnsistent messages. Comparisons were made disregarding topic-conditions, thus

1+1

permitting the determination of any topic-free trends in counter-attitudinal

and attitude-consistent encoding. These results are in Table 11.

 

 

 

 

 

 

 

Table 11. Intensifier Ratios g" “
(combined data) E
Required Message Output
Ratio Counter-Attitudinal Attitude-Consistent
I
High Polarity .361 .384 1
Modifiers per I
total modifiers
Low Polarity .132 .122
Modifiers per
total modifiers
High Polarity .598 .644*
Verbs per
total verbs
Low Polarity .146 .128
Verbs per
total verbs
*p < .05
The mean ratio differences in each of the four cases (as tested by means
of a correlated £_test) are in the predicted directions. One difference was
statistically Significant. When encoders used verbs, they used more highly
polarized verbs in their attitudinally-congruent messages than they did in
their counter—attitudinal messages. The mean differences are, however, all
consistent with the eXperimental predictions.
. V

42

Analysis of Variance Results

The use of the correlated 5 test to determine attitude-consistent vs.
counter-attitudinal differences may be conservative as perhaps permitting a
larger error term than the experimental design actually requires. Thus,
treatment by levels by replication 5 tests ("Case II" in Winer, 1962) were
done on each of the 24 dependent style and performance measures. The findings F
from this three factor analysis of variance with repeated measures are in I

Appendix 6 .

m—I-Lr n lia' ‘

The results of this additional analysis parallel the t-test findings.

 

With respect to the counter-attitudinal vs. attitudinal-consistent differences,
encoders were found to use relatively more verbs per total verbs used (p(.10)
and more passive voice verb forms (2 ( 10). No Significant differences were
found as a function of encoding order. Several topic differences were noted.
However, since there were no controls for degree of positiveness or negative-
ness of each issue, or for degree of encoder ego-involvement or topic salience,
interpretation of these topic difference findings cannot be conclusive. The

discussion chapter further discusses these latter results.

Individual Mood and Task Ratiggs

Two indices of subjects' individual reactions to the required encoding
tasks permitted comparisons of the attitude-consistent and counter-attitudinal
encoding situations: mood scale ratings and task evaluation ratings.

Mood Scale Ratinji. The differences between individual ratings of their
own mood or feelings when writing counter-attitudinal messages and their feelings
when writing attitude-congruent messages are listed in Table 12. For neither

tOpic condition did the differences in subjects' scores on any of the five

\1

v.3

mood factors (aggression, anxiety, deactivation, social affection, pleasant-
ness) approach Significance.

The data are presented in Table 12 in terms of mean factor scores for
subjects within each topic condition. A score of 3.00 would indicate minimum
perceived applicability of the adjectives within each factor and a score of

15.00 maximum perceived applicability.

 

Table 12. Mood Scale Ratings

 

Required Message Output

 

Message
Mood Factor Topic Counter-Attitudinal Attitude-Consistent
Aggression "18" 5.42 5.81
"TV" 5.84 5.63
Anxiety "18" 4.50 4.62
"TV" 4.90 4.47
Deactivation "18" 7.19 6.38
"TV" 7.32 7.32
Social "18" 5.88 5.92
Affection "TV" 5.37 4.74
Pleasantness "18" 6.96 7.19
"TV" 6.21 6.42

 

Topic Evaluation Rating_. Table 13 presents the results of subjects'
task pleasantness and difficulty ratings. Data are presented in terms of mean
scale ratings, on a five-point scale, within each topic condition. Higher mean
scores indicate more favorable task evaluations (i.e. , more pleasant, less

difficult).

 

nu

 

Table 13. Task Evaluation Ratings

 

Required Message Output

 

 

 

Evaluation Message
Scale Topic Counter—Attitudinal Attitude-Consistent
"unpleasant- "18" 3. 16 3. 60* ,. '
pleasant" -
"TV" 3. 5 3 3. 42
"di ffi cult- "18" 2 . 52 3. 24**
easy" ;
"'rv" 3.21 3.00 .1
* p (. 10
** p (. 005

 

For the 18 year old vote prOposition, as predicted, subjects rated the
task of writing an attitude-consistent message as significantly easier
(2 (.005) and somewhat more pleasant (p (.10) than they rated the counter-
attitudinal task. 7

On the federal TV censorship issue, there were no Significance differences

in subjects' task ratings.

CHAPTER IV

DISCUSSION

Summary

The present investigation attempted to assess the effects of cognitive
stress, induced by requiring attitude-discrepant communication behavior,
upon message output. Hypotheses were based on previous investigations into
the consequences of psychological stress on performance and from research
relating to language intensity as an indicant of author hesitancy or uncer-
tainty. It was generally predicted that a counter-attitudinal encoding
Situation would produce measurable deivations from an individual's custOmary
(attitude-consistent) encoding behavior in terms of an impairment of encoding
performance and a decrement in expressed language intensity.

Procedures. Experimental subjects were divided into two groups. One
group addressed a.proposition previously found to be a generally favorable one.
The second group encoded messages concerning a negatively evaluated proposition.
Each subject encoded two messages, one in opposition to his prior attitudes
and one in accord with those attitudes. The order in which the messages were
encoded was controlled for. There was no time limit placed on the task.

Comparisons of encoding performance and style variables were made'
between the two messages produced by each encoder. In addition, comparisons

were made of the subjects' personal mood and task evaluation ratings.

45

'“h

 

46

Message Production Measures. No significant difference was found between
an encoder's counter-attitudinal and attitude-consistent message production
rates. This was contrary to expectation. Subjects arguing counter-attitudinally
used no less time and encoded no more slowly than these same subjects arguing
in accord with their attitudes.

There was a tendency (p (.15) for subjects to make relatively more
message errors and alterations when writing counter-attitudinally. Although
the differences in numbers of errors made (punctuation, Spelling, editing,
total errors) per total words encoded did not reach acceptable levels of
significance, the direction in each case was in accord with predictions.

In addition, an examination of the rate of error productiOn (per minute
of time Spent encoding) showed an overall significantly greater rate of errors
for counter-attitudinally encoded messages.

Message Style Measures. Contrary to hypothesis, no significant differences
or consistent trends were observed for the message readability measures (words
per sentences and syllables per first hundred words). According to the measures
employed, subjects' counter-attitudinal isages were no less readable than
their attitude-consistent messages. \

Limited support was found for the language intensity hypothesis. In
terns of the polarity of the verbs used in each message, all analyses Showed
a consistent tendency for individuals arguing counter-attitudinally to some-
what use fewer high intensity verb forms.

There was also, for both tOpics and in all analyses, a tendency for subjects
to use relatively fewer active voice verbs (and thus more passive voice verbs)
when writing counter-attitudinal messages than when writing attitude-consistent

messages 0

 

47

No difference between counter-attitudinal and attitude-consistent
messages was found in the relative numbers of present tense verb fOrms used.

Analysis of the counter-attitudinal vs. attitude-consistent message
differences in percentages of high and low polarity modifiers per total words
and per total modifiers Showed consistent, though not Significant, trends in
the expected directions.

No Significant difference was found in the usage of ambivalence con-
structions as a function of subjects' agreement or disagreement with their
required message output.

Mood and Task Rating_: No Significant differences in subjects' rated
feelings or mood after encoding were observed.

For the positively-evaluated topic, Subjects, as expected, rated the
counter-attitudinal task as more difficult and somewhat less pleasant than
their attitude-consistent task. This difference was not found for subjects
writing messages about the negatively-evaluated.proposition.

Conclusions. From the data reported, it might be concluded that the
theory upon which the experimental hypotheses were based is in error. There
are, however, several points which would argue against this interpretation.

With respect to the perfOrmance measures, there is some indication that
the expectation of performance quality decrement as a function of counter-
attitudinally induced stress is warranted. Not only does the prior research
noted earlier consistently report such debilitation, the error measures
employed in the present investigation all seemingly evidence this general trend.
While not attaining statistical levels of significance, the consistency of the

findings seems noteworthy.

 

\‘L

48

The data do not provide support for the readability predictions. No
Significant differences in encoders' measured readability levels were observed
as a function of cognitive stress.

There does appear to be, however, at least some limited support for the
language intensity predictions. Again, rather than the actual size of the
differences, it is the consistency of the direction of these differences which "r "
bears mention. The language intensity predictions seem generally to have
been born out for the encoders' usage of verb forms. This support is not

evident for their use of modifiers and ambivalence constructions.

 

Unresolved Issues

Plausible reasons for the failure to completely confirm the experimental
predictions will be discussed here.

MethodologEal Shortcomings. Three potential areas where methodological
problems could have arisen are listed below: individual variance, stress
measures and style measures.

a. individual variance. One problem lies with the observed variance.
There was apparently some greater variation in individual style and per-
formance measures as a result of the increaSe in cognitive stress. As may be
seen from the standard deviations listed in Appendix 5, in 29 of 47 cases, the

variance among individuals arguing counter-attitudinally is larger than the
variance observed when these same individuals wrote attitude-consistent messages.
This increase in variance across all comparisons is statistically Significant

(p = .032, by Sign test). However, in only 2 of 48 such comparisons is there

significant heterogeneity of variance.

49

Thus, it is possible that the induction of stress increases the
variability in performance and style measures. That is, encoders may become
less organized and predictable in terms of their encoding behavior. This
increase in variance among counter-attitudinal encoders may mitigate somewhat
against the probability of obtaining significant differences. The general
trend seems to be in the direction of lower performance quality and decreased
language intensity. The increase in within-cell variatiOn, also seemingly a
result of cognitive stress, may however be masking the extent of this trend.

b. stress measures. One possible problem would seem to be the measure-

 

 

ment (or lack of measurement) of the stress induction. No direct manipulation
check was used, for several reasons. A more adequate check of the success in
inducing psychological stress would have attempted measurement immediately
after subjects received the counter-attitudinal message task. It was believed,
however, that such a measure might contribute more to subjects' perceptions of
the artificiality of the artificiality of the encoding situation than it would
to the measurement of stress. Because we wished subjects to write as "nor-
mally" as possible, and, since the time meaSure from reception of message
task instructions to completion of task would have been further complicated by
an intervening task, the existence of Stress was assessed only £123 message
production. .

That decision gives rise to several problems. The encoding process
itself may be somewhat stress-reducing, as suggested by Greenberg (1960) and
by Lynch (1963). Thus the post-encoding measurement could have been measuring

lesser amounts of stress than were actually created by the experimental task.

I

50

The instructions attempted to overcome this problem by asking subjects to rate
their feelings not as they now felt, but as they had previously felt, when
writing their assigned messages. Still, the possibility remained that there
was a reduction of stress occurring as a result of the message-encoding process.
This might account for the failure to find significant differences in the level
of post-message stress.

Of more general concern is the adequacy of the instrument itself. The
stress measure, originally suggested by Nowlis and Green (1957) and subsequently
used by Greenberg (1960) has not been otherwise widely used or accepted as a
measure of cognitive stress. This objection might be noted for any measure
proposed, however. There seems to be no adequate measure of the internalization
of "cognitive stress." The existence of stress is customarily inferred on the

basis of the stimuli introduced and the behavior change observed (e.g., Osgood,

1957; Lazarus et al., 1952). There is an apparent lack of consensually accepted,

reliable means of measuring stress.

c. message style measures. Further methodological considerations exist

 

in the manner in which the style of each message was quantified. There seemed
to be no way to avoid the necessity for some subjective decision-making, due
to the variables selected for consideration. However, the problem here does
not seem to be one of lack of reliability in the measures despite the inherent
subjectivity in the measurement of style. The problem rather might lie with

the alternative measures which could have also been used.

1'" .

 

..
5‘." 31." t

F!

Is "language intensity" best measured by polarity or intensity of modifiers
and verbs? What are the relative weights which should be assigned to each element?

Are verbs more important than modifiers as indicants of the uncertainty or

51

strength of feeling in a message source -- if so, how much more-2 What of the
nouns used by each encoder? No analysis was made of the nouns used, because
there appears to be no available scheme for noun classification in terms of
evaluative polarity.

Although the theoretic prediction of a counter-attitudinally induced decre-
ment in language intensity is clear, there seem to be only relatively indirect
measures of this intensity. Any number of different measures or ratios might
also have been used as perhaps being indicative of language polarity. Perhaps
the "concreteness" or "abstractness" of nouns (see Plesch, 1950; Gillie, 1957;
Haskins, 1960) bears some relationships to what this study has discussed as
language intensity. To the extent that concrete nouns are less ambiguously
interpreted by receivers, they may provide an additional index of the certainty
or opinion polarization apparent in an assertion.

Perhaps the degree of modification of nouns (adjectives/nouns) or verbs
(adverbs/verbs) or some combination might also serve as useful indices of the
hesitancy expressed in an author's statements. Boder's "adjective-verb
quotient" (1940) and Dollard and Mowrer's "discomfort-relief quotient" are
additional comparisons which might have been made. An examination of lexical
diversity (e.g. extent of common words or of function vs. content words, as
discussed by Rommetveit, 1968), a comparison of message "cloze" scores (as
proposed by Taylor, 1953) and/or a lock at the usage of personal pronouns
(suggested by Ruesch, 1953) would have been other possible considerations.

We chose only those measures with some prior empirical basis for expecting
a fairly direct relationship with language intensity. This does not rule out

the possibility of other, equally important message intensity indices. However,

 

-'- all-r .~..,. ..-.--.,,

3m

3"

52

due to the apparent lack of available research whiCh has systematically
linked message elements to receiver judgments of hesitancy, intensity or
strength, there has been no satisfactory determination of the most sensitive
measures of intensity.

Theoretical Problems. All hypotheses were made on the basis of pre-

 

 

dictions that a certain type of psychological stress (i.e. counter-attitudinally
induced stress) would produce Specific, measurable effects on an individual's

encoding behavior. The problem at this point is whether or not "stress" was

 

actually generalted by the experimental task requirements and, indeed, whether E ,5
stress is the only, or even the best, theoretic explanation for the observed
results. A I"

With respect to the first part of this problem (whether stress was actually
created), one is forced to argue from the obtained results. There appears to
have been a general performance decrement when encoders wrote in opposition
to their beliefS. In addition, for one of the two topics, the task evaluation
ratings showed subjects perceived the counter-attitudinal task as more difficult
and less pleasant. These findings would support the statement that there was
greater stress in the counter—attitudinal condition than in the attitude-
consistent condition. '

What would appear to be the basis for this stress? It seems reasonable
that subjects would know less about the existing arguments which Oppose their
onw attitudes. Although there were apparently no fewer total words encoded
by subjects when they wrote counter—attitudinally, there may well have been some
difference in the numbers or diversity of arguments employed or even in the
proportions of content vs. function words used. To the extent that one knows

less about what one is instructed to write on, some stress should exist.

 

 

53

It also seems that subjects would not be accustomed to writing counter-
attitudinally. Perhaps it is the novelty of the task requirements which accounts
for the change in writing behavior. This also may be a partial source of the
psychological stress posited in the present study.

Does an encoder tend to use relatively more low intensity words in a counter-
attitudinal Situation because of the Situational novelty or because of some lack FL.
of information about the other side? IS low intensity language a result of some
personal hesitancy, discomfort or uncertainty which one feels about writing

things he does not believe? Does an encoder write less intense messages

 

because stress situations cause one to be less likely to use strong or intense
language, or perhaps because of a tendency to encode Statements with which one
can more easily agree, as is suggested by "assimilation-contrast" theory (as
discussed by Sherif and Hovland, 1961)?

It does not appear possible to adequately separate out the more likely
of these causes from the less likely, at least in the present study. Perhaps
they are not actually separable, given the present measurement capabilities.
This may well be a problem for future research and measurement refinement.

For the present experiment, one can merely speculate as to the most probable

bases for the stress that was apparently generated.

Research Impli cations

 

Why study the counter-attitudinal encoding Situation? First, the
situation apparently has several "real-life" counterparts. It is not merely a
contrived experimental situation. PrOfessional encoders (editorial writers,
Speech writers, advertising copy writers) may be assigned a specific task of

writing messages with which they do not personally agree. It would seem logical

54

that such an occasion Should result in differences from an individual's more
normal encoding behavior. The present study begins to specify some of the
probable differences.

It appears that an encoder's prior attitudes influence how and what he
writes. The people who assign encoding tasks do not customarily take account
of the attitudes of the prospective encoders, however. To the extent that
professional encoders behave similarly to the way college business writing
students do (and this is a subject for. future inquiry), the present Study would
provide some support fot the contention that encoders' attitudes Should be
c onsidered. These attitudes should be considered, that is, if increased
errors and decreased intensity produce a less desirable message, one which
conforms less well to the hoped-for communication output.

Prior research indicates that messages of the type more likely produced
by encoders who do not agree with the output required of them would be less
effective. Greenberg and Razinsky (1966) found that increased numbers of
Spelling, punctuation and grammatical errors, at least at extreme levels,
resulted in decreased evaluations of message logic and clarity as well as
decreases in receiver ratings of source competence and trustworthiness. In
addition, the authors found decrements in the amount of information gained as
a function of high numbers of message errors.

McEwen and Greenberg (1969b) reported lower ratings of message clarity
and a generally lower evaluation of the meSsage source when receivers rated
messages employing low intensity language (low polarity verbs and modifiers).

Thus it would seem that the increased errors and decreased language intensity
observed as a result of counter-attitudinally induced stress would have some

Significant effects on the probable impact that these sorts of messages would have.

I! ELK-1'

 

55

A logical extension of the present study would be an investigation of
the effectiveness of counter-attitudinally encoded messages (as Opposed to
messages produced by attitude-consistent encoders) among groups of receivers.
Do the differences noted in message intensity make a Significant difference to
readers of that message? Does the difference in message effectiveness depend
upon the potential receivers of the messages? Perhaps audiences who are highly
opposed to the message proposal are more persuaded by the less intense messages
encoded under counter-attitudinal conditions (as in a "soft sell"). Highly
intense messages may be more effective among people who are in agreement with
or neutral toward the message proposition.

Further research could compare messages on the same Side of an issue,
differing only in whether or not the encoders who produced the messages agreed
with the message proposition. This might point out the extent to which the
observed message differences are meaningful differences, with respect to their
effects on attitude change toward source, message and topic as well as their
effects on other dependent measures (e.g. comprehension and retention).

Messages could also be compared in terms of exposure probabilities. To
the extent that language intensity correSpondS to the concept of stimulus
intensity, some effect on the amount of initial attention paid to messages in
fa voluntary exposure Situation would be expectedaf Again, there would be the
Opportunity to assess the meaningfulness of the observed message differences --
are the differences great enough to produce noticeable differences in the
probability that receivers will pay attention, either initially or over a
period of time, to a given message?

As mentioned befOre, much research has been devoted to thedktermination

of the factors influencing the amount of selfepersuasion fellowing belief-discrepant

 

56

communication behavior. Little has been done in the way of analyzing the
communication behavior produced. A Specification of the important message
variables reflective of individual attitudinal states Should provide correlates
which would aid in the investigation of attitude change. For instance, it
would be possible to examine the relationship between extent of attitude change
(i.e. self-persuasion) and encoding style change. The difference betveen an
encoder's attitude-consistent and counter-attitudinal message intensity levels
might provide a clue to the amount of personal inconsistency he is undergoing.
It may be that larger deviations from normal encoding styles are indicative of
greater dissonance and hence greater subsequent attitude change. On the other
hand, perhaps greater style deviations provide an alternative mode for the
reduction' of inconsistency or dissonance and thus less attitude change could
result. An investigation of this research area should thus allow for some pre-
dictions of amount of probable attitude change on the basis of observed message
style alterations.

The present study found Significant differences in encoding style and
performance characteristics as a function of the topic on which subjects wrote.’
As reported in Appendix 6, a number of Significant differences occurred between
the measures obtained from the encoders writing on the 18 year old vote issue
and those writing on the federal TV censorship issue. Due to the lack of
suitable controls, any conclusions here must be only tentative.

A number of the differences seem attributable to the particular nature of
the encoding requirements inherent in theSe two issues. For instance, subjects
made more Spelling errors and used relatively more syllables per hundred words

when writing messages addressing the federal television censorship proposition.

 

‘1

I _. E"

57

This seems reasonable merely as a result of "television censorship" being more
difficult to Spell and containing more syllables per word than "18 year old
vote." What is more interesting is the fact that encoders writing on the TV
censorship issue used more low intense verb forms, fewer ambivalence con-
structions, more future tense and fewer past tense verb ferns than those writing
on the 18 year old vote proposal.

The present study did not, however, attempt a systematic examination of
encoding differences as a function of message topic. This area seem a subject
for future inquiry. There did appear, in the present study, to be a relatively
greater tendency to conform to eXperimental predictions among encoders writing
on the "18 year old vote" issue. Assuming that this issue was more salient or
meaningful for college Students, some future research hypotheses might be
generated.

Further research might examine messages addressed to different issues,
saying in extent of'ego-involvement with either greater or lesser variability
in encoder attitude than was presently employed. One might look at differences
in messages arguing on mildly polarized or salient topics, as Opposed to the
language differences in messages addreSsing highly polarized or salient issues.
Thus, it might be possible to determine whether or not topic salience, for
example, has an effect on the degree to which individuals modify their encoding
performances when writing counter-attitudinally.

An examination of the act of writing a positive (proposition-favoring) as
Opposed to that of writing a negative (anti-proposition) message could also be
undertaken. How does the language intensity level, for example, of an encoder
differ when he is writing a negative message from what it is when he is writing

a positive message? It may be that individuals generally use more intense

 

58

language when writing messages of condemnation. Encoders may customarily
employ more polarized and emphatic language when they attack an issue than they
do when they praise an issue.

An examination of the encoding behavior of "neutral" people could also
be accomplished. To what extent are the relationships between agreement or
disagreement with required output and subsequent performance and style measures
linear? IS the language intensity used by a counter-attitudinal encoder less
than that used by a person who is neutral toward the message issue? Unfortunately,
in the present Study there were an insufficient number of encoders who were
neutral toward the two tOpics to permit adequate analysis of this problem.

Only through additional research will such questions be answered. It is
hoped that the present study at least has provided some impetus toward the

research necessary to answer them.

 

I...» '
|
.

References

Allport, G.W. and L.J. Postman The basic psychology of rumor. Transactions
af_the New York Acadeay 2: Science, 1945, 29 61-81.

 

Appley, M.H. and R. Trumbull Psychological Stress. New York: Appleton-
Century-Crofts, 1967.

Berger, C.R. The effects of influence feedback and need influence on the
relationship between incentive magnitude and attitude change. Unpub-
lished Ph.D. dissertation, Michigan State University, 1968.

Bettinghaus, E.P. and I.L. Preston Dogmatism and performance of the communi-
cator under cognitive stress. Journalism Quarterly, 1964, El, 399-402.

Boder, D.P. The adjective-verb quotient: a contribution to the study of
language. Psychological Record, 1940, is 309-343.

Carlsmith, J.M., B.E. Collins and R.L. Helmreich Studies in forced compliance;
I. the effect of pressure for compliance on attitude change produced
by face-to-face role playing and anonymous essay writing. Journal a£_
Personality aaa_Social Psychology, 1966, 39 1-13.

Clark, H.H. Some structural properties of Simple active and passive sentences.
Journal a£_Verbal Learning and Verbal Behavior, 1965, is 365-370.

Clevenger, T. Asynthesis of experimental research in stage fright. Quarterly
Journal 2f Speech, 1959, 33, 134-145.

Cliff, N. Adverbs as multipliers. Psychological Review, 1959, EB! 27—44.

Dodd, S.C. and T.R. Gerbrick Word scales for degrees of opinion. Langaaga
and Speech, 1960, B3 18-31.

Dollard, J. and O.H. Mowrer A method of measuring tension in written docu-
ments. Journal of Abnormal and Social Psychology, 1947, 53, 3-32.

Festinger, L. a_Theoay gf_§agnitive Dissonance. Stanford: StanfOrd
University Press, 1957.

Flesch, R. A new readability yardstick. Journal a£_Applied Psychology,
1948, 23, 221-233.

Plesch, R. Measuring the level of abstraction. Journal a£_Applied Psychology,
1950, 9:, 330-390.

Gillie, P.J. A Simplified formula for measuring abstractness in writing.
Journal 2: Applied Psychology, 1957, 223 214-217.

59

 

60

Greenberg, B.S. Performance and message consequences of encoding behavior
under cognitive stress. Unpublished Ph.D. dissertation, University
of Wisconsin, 1960.

Greenberg, B.S. and R.L. Razinsky Some effects of variations in message
quality. Journalism Quarterly, 1966, 42, 486-492.

Greenberg, 8.8. and P.H. Tannenbaum Communicator performance under cog-
nitive stress. Journalism Quarterly, 1962, 323 169-178.

Gynther, R.A. The effects of anxiety and of situational stress on communi- Ida
cative efficiency. Journal g£_Abnorma1 and Social Psychology, 1957,
51, 274-276.

Haskins, J.B. Validation of the abstraction index as a tool for content-
effects analysis and content analysis. Journal g§_Applied Psychology,
1960 , 41, 102-106 .

 

Holsti, 0. Computer techniques in content analysis: the measurement of
qualities and performance. Paper presented to the National Conference
on Content Analysis, 1967.

Howe, E.S. Probabilistic adverbial qualification of adjectives. Journal g§_
Verbal Learning and Verbal Behavior, 1962, 1, 225-242.

Howe, E.S. Verb tense, negatives, and other determinants of the intensity of
evaluative meaning. Journal 2: Verbal Learning and Verbal Behavior,
1966, 3, 147-155.

Janis, I.L. and J.B. Gilmore The influence of incentive conditions on the
success of role playing in modifying attitudes. Journal 2; Personality
and Social Psychology, 1965, 1, 17-27.

Johnson, M.G. Syntactic position and rated meaning. In L. Jakobovits and

M. Miron, Readin s in the Psychology of Lan a . Bnglewood Cliffs,
N.J.: Prentice-Hall, 1967, 472-479. "‘;ELJEi

Klapper, J.T. The Effects 2£_Mass Communication, New York: The Free Press, 1960.

Klare, G.R. The Measurement 2£_Readability, Ames: Iowa State University Press,
1963.

Lazarus, R.S., J. Deese and S.P. Osler The effects of psychological stress
upon performance. Psychological Bulletin, 1952, 42, 293-317.

Linder, B.E., J. Cooper and B.E. Jones Decision freedom as a determinant of
the role of incentive magnitude in attitude change. Journal 2£_Person-

ality and Social Psychology, 1967, 6, 245-254.

61

Lilly, R.S. The qualification of evaluative adjectives by frequency adverbs.
Journal g£_Verba1 Learning and Verbal Behavior, 1968, 1, 333-336.

Lilly, R.S. Multiplying values of intensive, probabilistic and frequency
adverbs when combined with potency adjectives. Journal g§_Verbal
Learning and Verbal Behavior, 1968, 1, 854-858.

Lynch, M.D. Modes of resolution of cognitive inconsistency and mood-stress
through communication activity. Unpublished Ph.D. dissertation,
University of Wisconsin, 1963.

 

 

McBwen, W.J. and 8.8. Greenberg Effects of communication assertion inten- F‘-|
sity on congruity principle predictions. Journal of Communication, .
1969 (in press). --

McBwen, W.J. and 8.8. Greenberg Message intensity effects on receiver
evaluations of source, topic and message. Paper presented to the
annual convention of the National Society for the Study of Communi- . ”
cation, 1969. E I

Nowlis, V. and R. Green Papers presented to the 65th annual American
Psychology Association convention, 1957.

Osgood, C.R. Motivational dynamics of language behavior. In M. Jones
(ed.) Nebraska Sygposium gn_Motivation, 1957, 348-424.

Osgood, C.R. The representational model and relevant research methods. In
I. Pool (ed.) Trends in_Content Analysis. Urbana: University of
Illinois Press, 1959, 33-88.

Osgood, C.R. Some effects of motivation on style of encoding. In T.
Sebeok (ed.) Style in_Langgage. Cambridge, Mass: M.I.T. Press,
1960, 293-306.

Osgood, C.R. and E.G. Walker Motivation and language behavior: a content
analysis of suicide notes. Journal 2: Abnormal and Social Psychology,
1959, 52, 58-67.

Rommetveit, R. Words, Meanings and Messages. New York: Academic Press,
1968.

Rosenberg, M.J. Some limits of dissonance: toward a differentiated view
of counter-attitudinal performance. In S. Peldman (ed.) Cognitive
Consistency. New York: Academic Press, 1966, 185-170.

 

Sherif, M. and C.I. Hovland Social Judggent. New Haven: Yale University
Press, 1961.

Siegel, S. Nonparametric Statistics. New York: McGraw-Hill, 1956.

62

Tannenbaum, P.H. The indexing process in communication. Public Opinion
Quarterly, 1955, 12, 292-302.

Taylor, W.L. "Cloze procedure": a new tool for measuring readability.
Journalism Quarterly, 1953, 32, 415-433.

 

Winer, B.J. Statistical Principles in_Experimental Design. New York:
McGraw-Hill, 1962.

 

«e

 

APPENDIX 1

TIME 1 ENCODING TASK INSTRUCTIONS

 

64

 

Please write your name in
this box:

 

Department of Communication

Study of F-

Persuasion and the Use of Arguments

 

 

A study is now being conducted at M.S.U. on the way people go abouts -~'

persuasing others. In order to help us determine how people use arguments,
we are asking each of you to write.what you feel to be a persuasive message.

We know that a person's attitude toward what he is writing about has
an effect on his use of arguments. So, we would like you to first fill out
the items on the next page which ask for your opinions toward some of the
topics you may be asked to write about.

Please complete the items on the next page and, when finished, S292.

so that we may begin the writing assignment together.

65

Please make a check mark on each scale according to how you personally
feel toward the statements listed below.

"The federal voting age requirements should be changed so as to permit
those 18 years of age and older the right to vote."

Wise : ° : ° ' : Foolish
Very Qu1te Slightly Equally Slightly Qu1te Very
Worthless : : : Valuable
very Quite H81ight1y Equally:81ightly Quite Very
Bad : : Good
Very Quite HSligEtly Equally Mlightly Qu1te Very
Fair : : : : - : Unfair
Very Quite Slighlty Equally Slightly Qu1te Very
Important : : : ° ° : Unimportant
very Qu1te Slightly Equally Slightly Quite Very
"Television programming should be censored by a federal control board."
Wise : : : : : ° Foolish
very Quite Slightly Equally Slightly Quite Very
Worthless : : : : : : Valuable
Very Quite Slightly Equally Slightly Qu1te Very
Bad : : : : ° : Good
Very Qu1te Slightly Equally Slightly Quite Very
Fair : : : : : : Unfair
Very Quite Slightly Equally Slightly Quite Very
Important : : ° ° ' : Unimportant
Very Quite Slightly Equally Slightly Qu1te Very

*******************inh'n'd:*1:*Q'czhhh’c************************************************

STOP.

finish this first section.
*sessseeseeseettee*esseteeeeeeseeeesessese*seseseeassessestsetseeteseeeseeeseeee

Do not go on to the next part until everyone has had an opportunity to

 

66

We would now like you to write a message which argues strongly IN EAXQR OE
the following statement: "The federal voting_gge requirements should be chaggeg.
so as to permit those 18 years of age_and older the right to vote."

We realize that you may or may not agree with the position you are asked to
take with regard to this issue. However, very often, eSpecially as new members in
an organization, you may be asked to write about things you do not completely
support. This assignment is similar to the request which might be made of you
if you were, for example, an advertising copy-writer or a newspaper editorial
writer.

Please write as strong and persuasive a message as you think you can. In
other words, we would like you to write a message, with as much emphasis as
possible, which could convince people that the minimum federal voting age should
be 18 years old.

You may write as much as you feel is necessary to produce a convincing message.
Also, take as much time as you need. We so not expect a perfect, finished copy
from you in the limits of this one class period. Therefore, feel free to make .
any changes or corrections in the message as you proceed. Don't worry about how
it looks, but what it ggyg,

You can help us by doing as good a job as you think you can. We wish to
thank you for your cooperation in this study.

Please RAISE YOUR HAND as soon as you have finished writing and editing
your paper and we will collect your papers and give you a short opinion-aire
to fill out.

Please begin writing.

 

 

i7

67

We would now like you to write a message which argues strongly AGAINST the
following statement: "The federal voting_2ge requirements should be changed so
as to permit those 18 years of age and older the right to vote."

We realize that you may or may not agree with the position you are asked
to take with regard to this issue. However, very often, especially as new members
in an organization, you may be asked to write about things you do not completely
support. This assignment is similar to the request which might be made of you
if you were, for example, an advertising copy-writer or a newspaper editorial

writer.

 

Please write as strong and persuasive a message as you think you can. In
other words, we would like you to write a message, with as much emphasis as
possible, which could convince people that the minimum federal voting age,should
323 be lowered to 18 years.

You may write as much as you feel is necessary to produce a convincing
message. Also, take‘as much time as you need. We do not expect a perfect,
finished copy from you in the limits of this one class period. Therefore,
feel free to make any changes or corrections in the message as you proceed.
Don‘t worry about how it looks, but what it 53331.

You can help us by doing as good a job as you think you can. We wish to
thank you for’your c00peration in this study.

Please RAISE YOUR HAND as soon as you have finished writing and editing
your paper and we will collect your papers and give you a short opinion-sire
to fill out.

Please begin writing.

 

68

We would now like you to write a message which argues strongly IN FAVOR OP
the following statement: "Television programming should be censored by a federal

control boar . "

 

We realize that you may or may not agree with the position you are asked to
take with regard to this issue. However, very often, especially as new members
in an organization, you may be asked to write about things you do not completely F
support. This assignment is similar to the request which might be made of you I.
if you were, for example, an advertising copy-writer or a newspaper editorial

wri te r.

 

Please write as strong and persuasive a message as you think you can. In

 

other words, we would like you to write a message, with as much emphasis as
possible, which could convince people that TV programming should be censored by
a federal control board.

You may write as much as you feel is necessary to produce a convincing message.
Also, take as much time as you need. We do not expect a perfect, finished-copy
from you in the limits of this one class period. Therefore, feel free to make any
changes or corrections in the message as you proceed. Don't worry about how it
looks, but what it m.

You can help us by doing as good a job as you think you can. We wish to
thank you for your co0peration in this study.

Please RAISE YOUR HAND as soon as you have finished writing and editing your
paper and we will collect your papers and give you a short opinion-airs to fill
out.

Please begin writing.

 

69

We would now like you to write a message which argues strongly AGAINST the
following statment: "Television_programming should be censored by a federal

control boad."

We realize that you may or may not agree with the position you are asked to
take with regard to this issue. However, very often, especially as new members
in an organization, you may be asked to write about things you do not completely E"fih
support. This assignment is similar to the request which might be made of you
if you were, for example, an advertising copy-write or a newspaper editorial ,

writer.

 

Please write as strong and persuasive a message as you think you can. In
other words, we would like you to write a message, with as much emphasis as
possible, which could convince people that Tv programming should.pp£_be censored
by a federal board.

You may write as much as you feel is necessary to produce a convincing
message. Also, take as much time as you need. We do not eXpect a perfect, finished
copy from you in the limits of this one class period. Therefore, feel free to
Inake any changes or corrections in the message as you proceed. Don't worry
about how it looks, but what it gayg,

You can help us by doing as good a job as you think you can. We wish to
thank you for cooperation in this study.

Please RAISE YOUR HAND as soon as you have finished writing and editing
your paper and we will collect your papers and give you a short opinion-aire
to fill out.

Please begin writing.

 

70

 

 

 

Your personal feelings can influence the manner in which you attempt
to persuade others. Since we are concerned with the way a person uses
arguments, we would like to know how you feel about the study you have just
participated in. We would also like to know your personal feelings or mood
as of this moment.

On the following pages are a number of items which ask for your own
feelings about the study and your participation in it. Please turn the page

and complete the items.

 

5‘.

71

Please fill in the items below. Make one check mark on each scale according
to how well you feel each adjective describes your mood or feelings when writing
your assigned message.

Definitely Somewhat Slightly Undecided Does not
Applies to Applies Applies Apply

MLmood

 

SLUGGISH

 

INSECURE

 

STARTLED

 

 

KINDLY

 

RBPRESHED

 

DEFIANT

 

DROWSY

 

PLEASED

 

APPECTIONATE

 

EBARPUL

 

ANGRY

 

LIGHTHBARTBD

 

TIRED

 

FORGIVING

 

 

 

REBELLIOUS J

 

 

 

 

‘.

Please turn the page and complete the items in the next section. 4"

72

I feel that the writing task which I just completed was:
___Very Pleasant
Somewhat Pleasant
___Half and Half
Somewhat Unpleasant

Very Unpleasant

I thought that the assignment was:
___Very Difficult
__Somewhat Difficult
_____Half and Half

Somewhat Easy

Very Easy

 

APPENDIX 2

TIME 2 ENCODING TASK INSTRUCTIONS

 

'9
I

74

 

NAME:

 

Persuasion and
The Use of Arguments [‘15:

(Part II)

 

You have recently participated in part of a study on the ways peOple

 

1117. ,_~._.
I
I

attempt to convince others.

As a further part of our study we would now like you to write another
message, in much the same way as you did before.

Please turn the page, read the instructions, and begin.

Thank you for helping us again.

75

A number of you were asked in Part I of this study to write messages with
which you personally did not agree. Some of you wrote messages with which you
dig_agree. To make allowances for these differences in your personal attitudes,
we would now like you to do the following.

We would like you to write a message which argues strongly IN FAVOR OF
the following statement: "The federal voting_age requirements should be changfl
so as to permit those 18_years of age and older the right to vote."

Please write asstrong and persuasive a message as you think you can. In
other words, we would like you to write a message, with as much emphasis as
possible, which could convince people that the minimum federal voting age should
be 18 years old.

You may write as much as you feel is necessary and may take as much time as
you need. Again, feel free to make any corrections or changes in the message as
you proceed.

You can help us by doing as good a job as you think you can. Please raise

jyour hand as soon as you have finished writing and editing your papers so that we

may collect them. Thank you.

Please begin writing.

 

 

76

A number of you were asked in Part I of this study to write messages with
which you personally did not agree. Some of you wrote messages with which you
'did agree. To make allowances for these differences in your personal attitudes,
we would now like you to do the following.

We would like you to write a message which argues strongly AGAINST the
following statement: "The federal voting agggqfimmnts should be changed so

as to permit those l8pyears of age and older the right to vote."

 

Please write asstrong and persuasive a message as you think you can. In

 

other words, we would like you to write a message, with as much emphasis as

 

possible, which could convince people that the minimum federal voting age should "m
go; be lowered to 18.
You may write as much as you feel is necessary and may take as much time as
you need. Again, feel free to make any corrections or changes in the message as
you proceed.
You can help us by doing as good a job as you think you can. Please raise
your hand as soon as you have finished writing and editing your papers so that
we may collect them. Thank you.

Please begin writing.

 

77

A number of you were asked in Part I of this study to write messages with
which you personally did not agree. Some of you wrote messages with which you
did_agree. To make allowances for these differences in your personal attitudes,
we would now like you to do the following.

We would like you to write a message which argues strongly IN £529§;95.
the following statement: "Television programming should be censored by a
federal control board."

Please write as strong and persuasive a message as you think you can. In

 

other words, we would like you to write a message, with as much emphasis as
possible, which could convince people that TV programming should be censored
by a federal board.

You may write as much as you feel is necessary and may take as much time
as you need. Again, feel free to make any corrections or changes in the message
as you proceed.

You can help us by doing as good a job as you think you can. Please raise
jyour hand as soon as you have finished writing and editing your papers so that
we may collect them. Thank you.

Please begin writing.

 

78

A number of you were asked in Part I of this study to write messages with
which you personally did not agree. Some of you wrote messages with which you
did agree. To make allowances for these differences in your personal attitudes,
we would now like you to do the following.

We would like you to write a message which argues strongly AGAINST the
following statement: "Television_programming should be censored by a federal

control board."

 

Please write as strong and persuasive a message as you think you can. In
other words, we would like you to write a message, with as much emphasis as possible,
which could convince people that TV programming should.pp£ be censored by.a
federal board.
You may write as much as you feel is necessary and may take as much time as
you need. Again, feel free to make any corrections or changes in the message as
you proceed.
You can help us by doing as good a job as you think you can. Please raise
your hand as soon as you have finished writing and editing your papers so that
we may collect them. Thank you.

Please begin writing.

 

79

 

NAME:

 

We would like to know your personal feelings about this part of the study.
Please indicate, on the items listed on the following pages, your reactions.

Thank you.

 

 

 

 

80

Please fill in the items below. Make one check mark on each scale according
to how well you feel each adjective describes your mood or feelings when writing
your assigned message.

Definitely Somewhat Slightly Undecided Does not

Applies to Applies Applies Apply
My mood

 

SLUGGISH

 

INSECURE

 

STARTLED

 

KINDLY

 

REFRESHED

 

DBFIANT

 

DROWSY

 

PLEASED

 

AFFECTIONATE

 

EBARFUL

 

ANGRY

 

LIGHTHEARTED

 

TIRED

 

FORGIVING

 

RBBELLIOUS

 

 

 

 

 

 

Please turn the page and complete the items in the next section.

 

81

I feel that the writing task which I just completed was:
____Very Pleasant
Somewhat Pleasant
___Half and Half
Somewhat Unpleasant

Very Unpleasant

I thought that the assignment was:
____Very Difficult
____Somewhat Difficult
_____Half and Half
_____‘_S omewh at Easy

Very Easy

APPENDIX 3

MESSAGE CODING INSTRUCTIONS

AND CODING FORMS

83

CODING INSTRUCTIONS

Word Count: Count all words in each message. Numerals (e.g. 1776, 23, .05)
each count as one word. Hyphenated words (jack-in-the-box) count as one
word, as do all connected initials (NBC, NET, NCAA).

 

Sentence Counp: Count as sentences only what the subject has grammatically
marked off (by periods, question marks, exclamation points) -- do not
interpret sentences.

 

S llable Count: Count numbers of syllables as they are (orally) pronounced.
When in doubt, sound the word out. Count numbers and initials as they
are pronounced (e.g. 18 is two; 70 is three; NIT is three).

VERBS

 

Do NOT code infinitive forms (e.g. to go, to be, etc.) or gerund forms (going,
being, having).

Active-Passive: Count as passive verb forms only those obviously in the
passive voice (X was hit by the ball; censorship has been opposed
by righteous people). All other forms are considered active.

 

Present-Past-Future: Count as PAST tense all verb forms, either simple or
compound which are obviously past tense (e.g. was, had given, opposed,
have been denied, have supported).

Count as FUTURE tense only those verbs accompanied by "will" or "shall"

 

(e.g. I shall remain against; they will not pass; but NOT they would come).

All other forms are considered PRESENT tense verbs.

VERB
INTENSITY:

 

HIGH: all unqualified indicative mood verb forms without auxiliary verbs

 

has

ound; was opposed)
also all imperative mood verb forms (look out, beware!)

also all obligatogy auxiliary verb forms of the following types:
must, have to, cannot, require, demand, insist, do.

(I gp; he _s_ay_s_; we find, I me, you have; but NOT used to go; tried _t_o help;

84

Coding Instructions (2):

MODERATE Intensity verbs: all qualified indicative mood verb forms
(including tenses other than the present when they have an auxiliary verb):
used to ; was ; have ; will ; had .

Also, all auxiliary verb forms of the following types: ought to;
would; should; plan to; mean to; like to. Also the following verbs:
allows/permits; induces; believe; feel; think.

LOW Intensity verbs: all conditional and subjunctive verb forms, such
as the following: can; could; might; may.

Also, verbs of the following type: seems to; appears to; tries/attempts
to; help to; suggest/propose.

 

NOTE: 1. verbs in clauses preceded by IF will all be coded as LOW
in intensity.

2. verbs in clauses preceded by PERHAPS or MAYBE will be
peereased in intensity one level (such that: verbs formerly
LOW will remain LOW; verbs formerly MODERATE will now be LOW;
verbs formerly HIGH will pp! be coded as MODERATE.

3. the particle NOT will not alter the intensity values of verbs
normally coded as HIGH in intensity. Verbs originally coded
as MODERATE will now be coded as HIGH; verbs originally coded
as LOW will now be coded as follows:

HIGH: can not; could not
LOW: might not; may not
MODERATE: All others originally coded as low.

MODI FIERS

Code all modifiers (adverbs and adjectives) which appear on the lists
below or may be considered as synonyms for the words on these lists.

HIGH: extreme 1y unquestionably
very , too undeniably
unusually definitely
decidedly certainly
adamantly doubt less 1y
total 1y obviously
comp lete 1y proven
entire ly surely

85

Coding Instructions (3)

HIGH positively all

(cont'd.) absolutely none
necessarily every
directly no one

nothing

repeatedly same/identical(ly)
always any
forever only
never/ever each
continually no
externally
perpetually
instantly
immediately

MODERATE:

—— so similar
quite most
pretty such
rather enough
much just
many on the whole
a number of adequately
almost

apparently
high greatly
small
large recently
few usually
little often
main rarely
seldom
now further
presently
soon also, too
not only
already

LOW INTENSITY:

slightly possibly
somewhat doubtfully
partially supposedly
some perhaps
a few probably
several maybe
fairly conceivably
one of... part of.. seemingly
among the . . . even
generally

, occasionally
eventually sometimes
subsequently 1nfrequently

ultimately indirectly

 

86

Coding Instructions (4)

AMBIVALENCE CONSTRUCTIONS: Code all conjunctions of the following types as
ambivalence constructions:
(also include any synonyms for these words)
but
however
although
even though
while (when used as a conjunction)
still (when a conjunction)
yet (when a conjunction)

ERRORS

PUNCTUATION: Count all omissions or incorrect usages of periods,
question marks (when obviously needed) as well as commas,
colons, quotation marks, failure to capitalize prOperly. Only
code those errors which you find readily apparent.

 

SPELLING: Count all spelling errors. When in doubt, consult the
fl 0 .
dictionary to determine Spelling accuracy.

EDITING: Count each cross-out of a single word or part of a word;
count each erasure of a word or a part of a word; count each
retracing of a word or letter(s) within a word; count each in-
sertion of a word or part of a word as an instance of Editing.

 

 

87

CODING FORM Counter-attitudinal Encoding Study

Coder

 

Subject Number

 

Time of Exper.: Tues.(Tl) Wed. (T2)

1. Total words

 

2. Time Spent

 

3. Total Sentences

 

4. Total Syllables (lst hundred words)

 

5. Total Verbs (Active)

 

 

 

6. Total Verbs (passive)

7. Total Verbs (Present Tense)

 

8. Total Verbs (Past Tense)

 

9. Total Verbs (Future Tense)

 

10. Errors a) total punctuation errors

 

b) total Spelling errors

 

c) total x-outs 8 insertions

11. Total HIGH intensity modifiers

 

12. Total MODERATE intensity modifiers

 

13. Total LOW intensity modifiers

14. Total HIGH intensity Verbs

 

15. Total MODERATE intensity verbs

 

16. Total LOW intensity verbs

 

17. Total Ambivalence constructions

 

MESSAGE ANALYSIS Counter-attitudinal Encoding

88

 

Subject Number

 

Time of Exper. Tues. (T1) Wed. (T2)

1.

10.

ll.

12.

13.

.114.

15.

16.

17.

18.

wds./minute

 

wds./sentence

 

syllables/first 100 words.

 

active verbs/active + passive verbs

 

present tense vbs./pres + past + fut vbs.
past tense vbs./pres + past + future vbs.

future tense vbs./pres + past + fut. vbs.

 

 

 

 

HIGH intensity Modifiers/ total wds.
MODERATE intensity Modifiers/ total wds.

LOW intensity Modifiers/ total wds.

 

 

HIGH intensity Verbs/ total wds.

 

MODERATE intensity Verbs/ total wds.

 

LOW intensity Verbs/ total wds.

 

Punctuation errors/total wds.

 

Spelling errors/total wds.

 

x-outs 8 insertions/total wds.

 

Total Errors 5 Changes/total wds.

 

Total Ambivalence Constructions/total wds.

 

 

APPENDIX 4

CODING RELIABILITIES

 

90

Inter-Coder Reliabilities

Reliability Check Reliability Recheck

 

Variable #1 (sample n=lO) # 2 (sample n=5)
Total Words .9993 --
Total Sentences .9991” --
Total Syllables (first

100 words) .8517 --

Total Active Verbs .9903 .9846
Total Passive Verbs .8910 .6054
Total Present Tense Verbs .9882 .9949
Total Past Tense Verbs .9761 .7647
Total Future Tense Verbs 1.0000 .9660
Total Punctuation Errors .5643 .8840
Total Spelling Errors .8872 .9220
Total Editing .9887 .8770
Total High Intensity Modifiers .9761 .9694
Total Moderate Inten31ty

Modifiers .8147 .9748
Total Low Intensity Modifiers .9254 .9146
Total High Intensity Verbs .8960 .9422
Total Moderate Intensity Verbs .8689 .9001
Total Low Intensity verbs .8769 .9588
Total Ambivalence Constructions .9254 .8164

APPENDIX 5

 

Variable

 

Production Rate

Time Spent
Punctuation Errors
Spelling Errors
Editing

Total Errors

Error Rate

Words per Sentence
Syllables per 100 wds

High Intense Modifs
(per total wds)

Moderate Intense Modifs
(per total wds)

Low Intense Modifs
(per total wds)

High Intense Modifs
(per total modifs)

Low Intense Modifs
(per total modifs)

High Intense Verbs
(per total wds)

92

Standard Deviations

Topic Condition

Counter-
Attitudinal

 

 

18
TV

18
TV

18
TV

18
TV

18
TV

18
TV

18
TV

18
TV

18
. TV

18
TV

18
TV

18
TV

18
TV

18
TV

18
TV

vote
censors

vote
censors

vote
censors

vote
censors

vote
censors

vote
censors

vote
censors

vote
censors

vote
censors

vote
censors

vote
censors

vote
censors

vote
censors

vote
censors

vote
censors

4.398
4.144

5.487
6.074

.017
.019

.009
.012

.038
.204

.044
.210

.521
.604

4.863
3.014

9.658
15.281

.013
.013

.016
.019

.009
.011

.150
.161

.102
.128

.019
.015

Attitude-
Consistent

 

3.594
4.768

5.346
6.431

.010
.017

.009
.018

.052
.063

.053
.078

.434
.456

5.334
3.911

12.300
13.230

.014
.016

.020
.013

.006
.009

.182
.239

.099
.089

.018
.012

 

Standard Deviations (con't)

Variable

 

Moderate Intense Verbs
(Per total wds)

Low Intense Verbs
(per total wds)

High Intense Verbs
(per total verbs)

Low Intense Verbs
(per total verbs)
Active Voice Verbs
Present Tense Verbs
Past Tense Verbs

Future Tense Verbs

Ambivalence Consts.

Topic Condition

93

 

18
TV

18
TV

18
TV

18
TV

18
TV

18
TV

18
TV

18
TV

18
TV

vote
censors

vote
censors

vote
censors

vote
censors

vote
censors

vote
censors

vote
censors

vote
censors

vote
censors

Counter-
Attitudinal

Attitude-
Consistent

 

 

.010
.021

.009
.011

.116
.161

.093
.114

.079
.073

.091
.099

.097
.071

.040
.086

.005
.005

.019
.014

.007
.009

.150
.124

.058
.079

.066
.048

.130
.072

.133
.058

.027
.054

.004
.003

 

APPENDIX 6

95

P Test Results

 

 

 

Dependent Variable Source of Variance Obtained F

Production Rate A (order effect) .1437
B (topic effect) .1952
C (encoding conditions) .8234
AB (order x topic) 2.0362
AC (order x enc.) .0019
BC (topic x enc.) .0107
ABC (order x topic x 1.5464

encoding)

Time Spent A .5367
B .1368
C .5386
AB 1.2479
AC 1.6964
BC .1174
ABC 6.8133 ***

Punctuation Errors A .0042
B 1.8113
C .7285
AB .0188
AC .1188
BC .0593
ABC .3355

Spelling Errors A .1312
B 4.1814 **
C .7785
AB .0155
AC .0906
BC .1748
ABC 2.6874

* p <,.10

** p ( .05

Wk p < .025

**** p < .01

96
F Test Results (con't)

Dependent Variable Source of Variance

 

 

Editing A
B
C

AB
AC
BC
ABC

Total Errors

0313’

Error Rate

Words Per Sentence

Syllables Per Hundred A
Words B
C

Obtained P

 

2.8801
.9361
.7229

.6107
1.4639
1.0827

1.7682

2.5241
1.6857
1.2781

.4949
1.6892
1.0127

1.2900

3.8974 *
1.7957
2.6001

.4963
2.0099
.2028

.0432

.7513
4.5482 **
.1316

.1133
.0318
.3065

1.0584

2.1913

5.9540 **
.5290
.6155
.5972
.0008

3.3730

 

P Test Results (con't)

Dependent Variable

High Intensity Modifiers,

(per total words)

Moderate Intensity
Modifiers
(per total words)

Low Intensity Modifiers
(per total words)

High Intensity Modifier
Ratio
(per total modifiers)

Low Intensity Modifier
Ratio
(per total modifiers)

97

Source of Variance

 

I
A ,1 ~
B .
C

AB
AC
BC

ABC

Otflfi’

AB
AC
BC

ABC

ow>

as; °w>

g

C

Obtained F

 

.0077
2.1011
.2865

.0077
2.6477
.5360

1.9660

.0214
3.7325 *
.4526

.0249
3.3769 *
2.1873

1.8861

.0270
.2634
.0018

.1695
.0021
.6441

.1217

.0939
3.6973 *
.8788

.0026
.2489
.3609

3.5269 *
.2366
.2573
.2387
.0036
.3906
.3993

.0702

F Test Results (con't)

Dependent Variable

 

High Intensity Verbs
(per total words)

Moderate Intensity Verbs
(per total words)

Low Intensity Verbs
(per total words)

High Intensity Verb
Ratio
(per total verbs)

Low Intensity Verb
Ratio

98

Source of Variance

sag °w>

p.

BC

061:»

8‘86

Obtained F

 

.0130
.1534
.8206

.5704
.0168
.3932

4.1317

.1221
.0048
1.0058

.0052
.6907
1.7236

2.2977

.4807
8.7913
.4303

1.5860
.0887
.2918

.0781

.2760
1.6234
3.3796

1.1636
.3945
.3968

3.0535
.4660
7.9834
.8120
1.5996
.0843
.2440

.1264

****

*tth

F Test Results (con't)

 

Dependent Variable

Active Voice Verbs

Present Tense Verbs

Past Tense Verbs

Future Tense Verbs

Ambivalence Constructions

99

Source of Variance

 

8833

ABC

nun»

BC

ABC

033:»

OD!)

8‘86

ABC

Obtained P

 

.1654
.0449
3.1459

1.4345
2.1276
.4579

.1869

.5136
4.7916
.0261

.0249
.2015
2.4957

.2849

.0142
8.8698
.4757

.0998
.5664
1.5273

.2260

.4304
4.2123
.4215

.2903
.0684
.8013

.1086
2.6710
6.4204

.0369

.7160
3.0767

.3647

1.8018

it

*hth

it

***

*

 

“7'74? 1141474117444 @743