THE MEANING AND RANKEHG
OF VALUES

Thesis {for the chm of M. A.
MICHIGAN STATE UNIVERSITY

Robert J. Homant
1967

 

ABSTRACT
THE MEANING AND RANKING
OF VALUES
by Robert J. Homant

The purpose of this research was to investigate several ques-
tions about the meaning of Rokeach's terminal and instrumental
values (Rokeach, 1967). A value is defined as follows:

To say that a person has a value is to say

that he has an enduring belief that a specific
mode of conduct (an instrumental value) or end
state of existence (a terminal value) is person-
ally and socially preferable to alternative modes
of conduct or end states of existence.

"Meaning" refers to either connotative meaning (which is opera-
tionally defined by Osgood's semantic differential) or to associa-
tive meaning.

This research was addressed to the following four questions:

1) Does the rank ordering of the values accurately reflect
the subjects' evaluation of these concepts, as measured by
the semantic differential?

2) What is the relation of the value ranking to the other
(non-evaluative) dimensions of the semantic differential?
3) Are there any systematic differences in the associative
meaning of a particular value for a group of individuals
if some rank that value high and others rank that value
low?

4) To what extent can we say that values mean the same '

thing to different people regardless of how they rank them?

Robert J. Homant
In order to answer this fourth question, a set of control
words was constructed. Thus the question to be answered
was whether people agreed on the meaning of values at
least as much as they agreed on the meaning of non-value
words of the same frequency of usage.

264 83 defined the meaning of the values and the set of control
words, either with the semantic differential or by giving single
word associates or short phrases that would "show what these words
mean to you."

The evaluative factor of the semantic differential was found to
be highly correlated with value preference on Rokeach's terminal and
instrumental value scales, while the potency and activity factors
were only slightly related to the values. No difference was found
in the associative meaning of the individual values as a function of
their rank. The 83 showed agreement on the connotative meaning of
both the terminal and instrumental values. The $3 tended to show
agreement on the associative meaning of the instrumental values but
not the terminal values.

The only negative implication for the use of the value scales
concerned the associative meaning of the terminal values. It was
suggested that whenever there is concern with individual differences
in value preference (especially when the focus is on only a few of
the terminal values), care should be taken to measure the associa-
tive meaning of these values. Other possible wasy to measure this

meaning were suggested.

7 ,
yjf‘tvxfi-Jcé- ~ CJW4 / 7, /743"

, ’7
l ' / , n ’
) A LCZ 011“ /, cyflaa—ad

THE MEANING AND RANKING
0F VALUES
By

Robert J. Homant

A THESIS

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

MASTER OF ARTS
Department of Psychology

1967

(_ 9 ..

r A p

ACKNOWLEDGMENTS

Thanks are due to my committee, Drs. Bertram Garskof and James
Uleman, and especially to the chairman, Dr. Milton Rokeach, on whose
work this thesis is based.

Thanks are especially due to my typist, Miss Judy Hale, without

whose courageous typing this thesis may never have been forthcoming.

ii

CONTENTS

INTRODUCTION.OOCOOOOOOOOOOOOOOO0......O.O...OOOOOOOOOOOOOOOOOOOOOO. 1

Theoretical Positions on Value

The Definition and Measurement 6f Value

The Development of Self-Report Measures of Value
The Measurement of Meaning

Obtaining Associative Response Hierarchies

The Similarity of Associative Hierarchies
Application of Measuring Techniques to Values
Use of Meaning in this Study

Purposes of this Study

MHODOOOOOOOOOOOOOOOOOOO.0.00....OOOOOOOOOOOOOOOOOOOOOOOOOOO.O...

Subjects

Initial Procedure

Part A: Associations to Terminal Values

Part B: Associations to Instrumental Values

Part C: Semantic Differential Ratings for the Terminal Values

Part D: Semantic Differential Ratings for the Instrumental
Values

Obtaining Control Words

Part E: Baseline Data for Associations

Part F: Baseline Data for Semantic Differential Ratings

Deletion of Values' Explanatory Phrases

Part G: Association to Terminal Values without Explanatory
Phrases

RESULTSOOOOOOOOOOOOOOOOOOOOOOOOOCOOO...OOOOOOCOOOOOOOCOOOO00......

Check of the Semantic Differential Factor Structure
Scoring the Associations

Computations with the Semantic Differential

Normative Data on the Associative Response Hierarchies
Results Specific to Question 1

Results Specific to Question 2

Results Specific to Question 3

Results Specific to Question 4

DISCUSSIONOOOOOO00.0.0.0...OOOOOOOOOOOOOOOOOOOOOOOOO00.000.00.000.

Questions 1 and 2

Question 3

Question 4

Implications for Use of Value Scale

iii

1
4
8
12
13
14
16
19
19

23

23
23
24
25
26

27
27
29
29
30

31

31

31
35
41
42
42
47
50
51

56

56
61
63
65

REFERENCESOOOOOOOOOOOOOOO...OOOOOOOOOOOOOOOOOOO...OOOOOOOOOOOOOOO. 68
APPENDIX AC..0...0.0.0....CO.O.OOOOOOCOOOCOOIOOOOOIOOOOOOOOOOOO0.. 71

Terminal and Instrumental Values Used on Rokeach's Value
Scales 71

APPENDIX B.0.0000000000000000.0.00.00.00.00...OOOOOOOOOOOOOOOOO... 8‘12

Median Ranks Assigned to Values by Subjects for This
Experiment 72

APPENDIX C. C C O C. C . C O C O C O O O C C O O O C O O C O O C O O O O C C O O O O O O O O O O O O O O O O O O O O O O O 3‘4
Semantic Differential Instructions 74
APPENDIX DOOOOOOOOOOOOOOOOOOOO0...OOOOOOOOOOOOOOOOOOOOOO0.0.000... 76

Ten most Frequent Associations Given for Each Value 76

iv

10.

11.

12.

13.

LIST OF TABLES

Three-factor verimax solution of the 18 semantic differ-
ential scales used in this study.............................33

Mean correlation of the three sets of five scales with
the three faccors...OOOOOOOOOOOOOOOOOOOO0......0.0.00.000000034

Mean correlation of the Three sets of two scales with the

three faCtorSOOOOOOOOOOCOO...0.00.0000...0.0.00.000000000000036

Mean MR (measure of relatedness) for terminal values, for
total group, upper quartile, between quartile, and lower

quartileOOIOOOOOO00.000.000.00...0.0.0.0...OOOOOOOOOOOOOOOOOO38

Mean MR (measure of relatedness) fur instrumental values,
for total group, upper quartile, between quartile, and
lower quartile........................................o......39

Mean and variance for terminal values on three semantic
differential factors for total group, upper quartile,
and lower quartileOOOOOOOO...0.0.0.0000...0.00.00.00.0000000043

Mean and variance of instrumental values on three seman-
tic differential factors for total group, upper quartile,
and lower quartiIE......................o....................45

Range and median correlations between terminal and in-
strumental value scales and three semantic differential
faCtors-’15 scale semantic differentials00.000.000.00so0000.948

Range and median correlations between terminal and instrument-
al value scales and three semantic differential factors--6
Scale semantic differential.oooooooooooososo.ooooooooooo0000049

Summary table for the analysis of variance testing for
the trend of the mean MR for terminal values.................52

Summary table for the analysis of variance testing for
the trend of the mean MR for instrumental values.............53

Mean and variance of control words (given in order of
presentation) on three semantic differential factors.........55

Mean MR (measure of relatedness) for terminal values,
Without defining phrases.....................................57

14.

15.

16.

Mean MR (measure of relatedness) for instrumental values,
WithOUt defining pllraseSCCCOOOOCOOOOOCOOCOCCOCOOOCOOO00.00.0058

Mean MR (measure of relatedness) for control words...........59

Summary of.£ tests comparing values to control words, for
associative response hierarchies and semantic differential...60

vi

LIST OF ILLUSTRATIONS
Figure

1. Schematic diagram of the distribution of MR's for
any partiCUI-ar value............COCOCOOCOOCO00......00000040

vii

INTRODUCTION

A review article by Dukes (1955) points out many of the reasons
why the study of values has been a recurrent topic for psychologists.
Values have been shown to relate to several other variables, such as
sex, religious preference, occupation, age, academic achievement, etc.
(e.g., Woodruff, 1942, and Duffy, 1940). However, there have been
problems both of definition and measurement.

The purpose of this thesis will be to investigate the meaning
of values. Specifically, it will be concerned with the values which
Rokeach (1967) has included on his terminal and instrumental value
scales (see Appendix A). Before adopting a definition of value,
however, we will present a sample of the many positions on value
in order to provide a background to the problem of definition. In-
stead of a chronological presentation, the following positions have
been ordered to represent a range in theoretical viewpoint from.phenomr
enological to behavioral.
Theoretical Positions 93 11195

Burt (1963) argues that psychology should study value for its
own sake. He does not feel that "value" is a subjective phenomenon,
but that it has concrete properties which are discoverable to the
scientist. As an example of his position on value, he says that he
feels that an aesthetic value (e.g., beauty) has properties of its
own which are independent of any experiencing subject, and that an
object may be called beautiful in and of itself. He cites some experi-

mental evidence on the objectivity of value judgments which is supportive

2
of his position. It may be noted that most psychologists would
consider this approach entirely subjective, although Burt himself
would claim objectivity.

Another approach to value which shares Burt's contention that
psychologists should study value for its own sake is put forth by
Winthrop (1960, 1961). Winthrop, however, asserts that he uses
a phenomenological approach, and also differs from Burt by holding
that value is an intrasubjective phenomenon. He equates value with
what Burt would call the experience of value. Although Winthrop
feels that subjective experience cannot be measured quantitatively,
he thinks that psychology can get at some of the structure of values.
He recommends ordinal measurement for this task. Furthermore, he
believes that in experiencing value, there can be a discrepancy within
the human organism between value as a cognitive map (or a belief about
what is valuable), and value as an affective action tendency. This
discrepancy, he asserts, is a relevant concern for the psychotherapist's
understanding of neurosis.

A non-phenomenological, but still cognitive approach to value is
exemplified by Piaget (1932, 1962). For Piaget, a value is analogous
to a cognition. It differs from a cognition in that it is an affective
concept about what is good or desirable. Consequently it functions as
a standard of judgment for an individual's behavioral choices. Piaget
explains that to the extent that an individual has arranged his values

hierarchically, he will be relatively untroubled by conflict situations.

I have termed this approach non-phenomenological (in contrast to

Winthrop's) because Piaget is not.concerned with the subjective experience

3
of value. Rather he is concerned only with the effect that values
have on behavior in conflict (or choice) situations. However, Piaget's
anecdotal approach and "common sense" illustrations do not put him on
nearly as solid an operational footing as the two examples to follow.

Hill (1960) presents a theory in which the psychologist treats
values as inferences which he makes from overt behavior (including
verbal reports and paper-and-pencil testing). Hill's aim is to explain
the acquisition of values through an extension of (basically Hullian)
learning theory. He hypothesizes that children's learning of imitation
includes learning to conform to abstract ethical exhortations. Behaving
in accordance with a particular parental exhortation would then become
largely a matter of concept learning.

Although he uses what is basically the same learning theory orien-
tation as Hill, Martin (1954) maintains that values are learned by imita-
tion of behavior, and that parental exhortations are irrelevant if the
parents' own behavior is discrepant from them. For instance, if a
father should urge his son to be ambitious, and yet display only
laziness or lack of ambition himself, the son is more likely to acquire
lack of ambition (insofar as his father is a model for imitation). If
values are to become labeled (conscious), says Martin, they must be in-
ferred by the individual from.his own behavior--or this inference must
be made by others if the value remains unconscious.

Thus Hill and Martin seem to be in almost direct disagreement as
to whether the concept of a particular value or the behavior which

accompanies that value is learned first. Besides differing in

4
predicting what values would be learned, Martin is in the position of
being able to maintain that there are unconscious (or unlabeled)
values, whereas Hill is not.

The five positions on value which we have just reviewed should
not be taken as an exhaustive list by any means. Rather they are a
sample of the range of possible theoretical positions. Some other
uses of the term."value" can be found, for example, in the work of
Helson (1953), Lecky (1945), and Tolman (1951).
13 Definition and Measurement of“ Value

When one considers these diverse approaches to what values are,
it is not surprising that the measurement of values has had a contro-
versial history. McCurdy (1950) lists four general techniques by
which values (loosely defined as the strength or importance of

interests) can be measured:

1) self-report, including a wide variety of paper-and-
pencil techniques;

2) intensity of emotional reaction;
3) variations in observed moral judgment;

4) choice of "fealty" (for example, whether a person
engages in business or golf).

However, note that the last three of these measures require the in-

vestigator to observe the subject's behavior. They begin by assuming
that values are reflected in behavior. Therefore, they are unsuitable
techniques if one wishes to investigate the influence of values (defined
in terms of cognitive structure) on that behavior.

All four techniques measure something slightly different, and all

four attempt to measure value. This indicates that whenever psychologists

5
talk about value, a definition is very much needed. It is to this
problem that we now turn.

Barton (1962) summarized the different ways in which value has
been treated within an objective framework. His discussion points to
the source of the ambiguity in the use of the term "value." The different
uses of "value" can be subdivided into four major classifications: (l)
explicit preferential, (2) implicit preferential, (3) explicit normative,
and (4) implicit normative.

The distinction between explicit and implicit refers to whether
or not the psychologist infers the value from the subject's behavior
(an implicit value) or whether he treatsthe subject's verbal report
as indicative of or equivalent to a value. Naturally this break-
down is a matter of degree. The Allport-Vernon-Lindzey Study of
Values exemplifies a method that is partly explicit and partly
implicit. On this test, values are inferred from a series of
attitude and interest questions. Thus a value is inferred from
an explicit report on matters that are presumed to reflect the
value.

Regardless of whether a value is explicit or implicit, it
may also be classified as either "preferential" or "normative." In
general, preferential values are goals which the individual has-~things
he "prefers" to seek. Normative values are qualities used for judging.
For instance, a person may judge others (or himself) as good or bad

according to whether or not they are honest. However, if a person

6
thinks that children-~but not adults--should be obedient, then the
question of whether this person values obedience (normatively) can
be given two different answers.

However, the distinction between normative and preferential
values can become ambiguous. This may happen when an individual
feels that normative values (qualities used for judging) are virtues
or habits to be acquired. For example, he may seek to acquire
honesty or efficiency, and these qualities will function as preferenr
tial values for him.

Besides this type of distinction within each of the classes,
there is also a question of extension. For instance, within the
explicit preferential class, one (e.g., Campbell, 1963) can say that
any object or state-of-being that a person desires is a value, thus
making the concept equivalent to any "reported valence." On the other
hand, one can say that a value must be a relatively permanent state of
being which a person not only desires but also cognitively evaluates
as good.

Since the problem being investigated in this thesis is the
meaning of the values that are used by Rokeach (1967) on his value
scales, we adopt Rokeach's definition of value for this paper. Ro-
keach defines a value as follows:

To say that a person has a value is to say that he has
an enduring belief that a specific mode of conduct or
end-state of existence is personally and socially pref-
erable to alternative modes of conduct or end-states of
existence.

In terms of Barton's classification, this definition is explicit,

both normative and preferential. However the normative part is

7
limited to a person's judgments about how he himself should behave.
What Barton has termed "normative values" are referred to by Rokeach
as "instrumental values," and "preferential values" are referred to
as "terminal values." This usage will be maintained throughout this
paper. Since what we are now concerned with is a beliefl, we will
review here the use of self-report (the first technique on McCurdy's
list) to measure values.
Barton (1962) lists four types of paper-and-pencil measures
which.mmy be used to measure values:
1) forcedpchoice interest type (like the Allport-
Vernoanindzey);
2) self-prediction of one's probable behavior in a
hypothetical situation;
3) analysis of verbal reports in different situations
(e.g., White, 1951);
4) direct ranking or rating of values.
The first three of these techniques measure value as defined
in this paper only indirectly. Each of them requires an inference

2

on the part of the experimenter-~for example from.an attitude measure

 

1Rokeach (1967) defines a belief as follows: "A belief is any simple
proposition, conscious or unconscious, inferred from what a person
says or does, capable of being preceded by the phrase 'I believe
that...'" At least to the extent that beliefs are conscious, there-
fore, it follows that the most direct way to measure them is simply
to ask the subject what he does believe (via some type of self-report

instrument).

2Again following Rokeach (1967), an attitude is defined as, "A relatively
enduring organization of beliefs around an object or situation predis-
posing one to respond in some preferential manner." Thus the first

three of the self-report measures listed above are said to be direct
measures of attitude rather than value because they ask for a report

of the subject's behavior in specific situations.

to a value.

An example of this is the previously mentioned Allport-
Vernoanindzey Study of VAlues, which gives an individual a profile
in terms of six values (economic, aesthetic, political, religious,
social, and theoretical). The test is based on a series of attitude
and interest questions which can only be presumed to reflect the
value (see Gage, 1959).

The fourth technique, using rating or ranking scales, has some
slight precedent in the work of Wickert (1940), Woodruff (1942), and
Smith (1949), and has been used in different forms by Morris and

Jones (1955), Rosenberg (1956), and Rokeach (1967).

4 .1112 W 2!. W “flames 9.: ____Va1ue

Wickert (1940a) devised a test to measure nine goal-values:
freedom, helpfulness, new experience, power, recognition, response
(friendship), security, and workmanship. These values were chosen
by Wickert after analyzing 73 separate lists of values, and the
resulting list of nine values represents an advance over the Allport-
Vernon test which was based mainly on Spranger's (1928) six character
types. The measurement of the nine values, however, was only slightly
some direct than on the Allport-Vernon test; Wickert used an lB-item
multiple choice questionnaire, with four possible responses for
each question (each response was keyed to a value). The advantage
over the Allport-Vernon Study of Values was mainly in that the

content of the choices more closely reflected the value being measured.

9
However, Wickert's test still tried to measure an abstract (or
general) value in specific contexts.

WOodruff (1942) devised the "Study of Choices" by which he could
obtain the rank order of a person's value. The method was essentially
similar to Wickert's, in that it called for the respondent to make
a choice from alternative courses of behavior in response to some
hypothetical situation. As far as format goes, this test is not
much different from Wickert's test. However the choices are slightly
more general, and the value being measured is more directly reflected
in them, so that it approaches the idea of simply choosing values.
lWoodruff, for example,asks the subject to choose between joining one
of several groups. If the subject chooses the group which is
"closely related to a religious organization and encourages religious
ideals and sociability" the subject is said to have preferred the
value "Religion."

Smith (1949) employed a direct measure of what people value in
a technique that was designed for interviewing. The method simply

required the interviewer to ask the respondent what he felt was most
important in life. The answer wasthen coded under such headings as
economic security, home and family life, liberty and freedom, etc.
Then Morris and Jones (1955) used a seven-point Likert scale
for rating preferences toward each of 13 values which Morris had
developed (1956). These values were not defined as simple concepts,
but were beliefs about "ways to live," which were defined in exten-
sive paragraphs. For example, Morris' first "way to live" reads

as follows:

-10-
In this "design for living” the individual actively

. participates in the social life of his community, not
to change it, primarily, but to understand, appreciate,
and preserve the best that man has attained. Excessive
desires should be avoided and moderation sought. One
wants the good things of life but in an orderly way.
Life is to have clarity, balance, refinement, control.
Vulgarity, great enthusiasm, irrational behavior, im-
patience, indulgence are to be avoided. Friendship is
to be esteemed but not easy intimacy with many people.
Life is to have discipline, intelligibility, good man-
ners, predictability. Social changes are to be made
slowly and carefully, so that what has been achieved
in human culture is not lost. The individual should
be active physically and socially, but not in a hectic
or radical way. Restraint and intelligence should give
order to an active life.

Subjects simply placed checkmarks on the Likert scales which were beneath

each of the "ways to live."

Rosenberg (1956) also measured values directly by having subjects
sort thirty-five cards into twenty-one categories from "gives me maxi-
mum satisfaction" to "gives me maximum dissatisfaction." After they
had completed the sorting task, subjects ranked any values that had
been placed within the same category, thus generating a rank order of
all 35 values. Although this test had reliabilities of .89, Rosenberg
was primarily interested in the affective relation of attitudes toward
some of the 35 values, and little further work was done with this
technique.

Finally Rokeach (1967) simply presented subjects with two lists
of 12 values each, and had them rank order the values from 1 to 12 in
order of importance. The first list of values contained such terminal
goals as: a meaningful life, a comfortable life, a world at peace,
equality, freedom, etc. The second list contained such instrumental

values as: broadminded, clean, cooperative, courageous, forgiving, etc.

In the interest of reliability and comprehensiveness the original value

-11-

scales have been revised so that now there are two sets of 18 values,
each of which has an explanatory word or phrase added to it (see Appen-
dix A).

The ranking technique as such has the advantage of forcing the
subject to generate a value system. This is important if one hypothsizes
that behavior is determined by the relative (rather than the absolute)
importance of a person's values.

The use of simple concepts has the advantage of simplicity,
especially when compared to the long descriptive phrases of Morris'

"ways to live."

At the same time, however, this simplicity raises a new problem--that
of the meaning of the values to different subjects. Wickert (l940b)--in
discussing his own test--expressed the concern that "The actual verbal
content obviously contains different meanings to the different subjects."
Wickert reports that his test underwent considerable revision in order
to eliminate this unwanted variability. It should be noted that Wickert
is only asserting that the meaning of the verbal content varies between
subjects. Furthermore he seems to imply that this variability is greater
than it would be for a test not concerned with values.

This same type of viewpoint is also evident in a different context
in the theoretical framework of Dolland and Miller (1950). In their
work they assert first of all that verbally mediated motives (which
include what we have called values) can be powerful secondary drives.
They give the example that labeling a type of behavior "dishonest" can
be a very powerful stimulus to a person. They go on to qualify this,

however, by asserting that a concept like "honesty" means different things

-12-
to different people, and therefore we cannot expect to predict a person's
behavior simply by knowing how high a value he places on "honesty."

Of course the problem of whether words mean the same thing to
different people is basic to all science. Carnap (1934) points out
that in the final analysis all that one can hope to do within a science
is to define things in terms of a protocol (or fundamental) language
which must be assumed to communicate knowledge between individuals. That
is, the meaning of a word is not something that can be measured, since
ultimately the measuring instrument itself would only become another form
of the concept.

In the psychological literature, Noble (1952, 1963) best makes this
same point by explaining that the "meaning" relationship (when used
verbally) is transitive, symmetrical, and reflexive, and is therefore
intrinsically non-measurable.

The Measurement 2£_Meaning

'Noble, however, did not feel that this meant that the area of meaning
should be abandoned by psychology. Rather he turned his attention to
measuring "meaningfulness." Stated simply, since what a concept means
grammatically is determined by the relationships it can have with other
concepts, then any two concepts can differ among themselves in the amount
of meaning (or meaningfulness: the number of relationships which they
evoke) that they have. The meaningfulness of a word is a funcdon of the
number of associations which a person has to that word, and a word may be
said to "mean" (in some sense) the responses that it elicits.

Meaningfulness (symbolized by m or t_n_') has been measured by Noble

in two ways. The average m,of a word is defined as the "average number

:13.

of continued written associations made by a representative sample of
subjects during a standard time interval (60 seconds)." The average
3} is simply the average of the estimates of the number of associations
which subjects feel that they can give to a word. These two measures
correlate .918 for group data (Noble and Parker, 1960).
Obtaining Associative Response Hierarchies

In line with this type of reasoning, Deese (1965) refers to
"associative meaning." Two concepts have the same associative meaning
to the extent that they elicit the same distribution of associations.

Several techniques exist for obtaining the associations to a concept.
Noble (1952) takes all the free associations to a word that a subject
can give in one minute. Deese (1965) allows only one response per
subject, and points out that obtaining multiple responses from the same
subject brings up the problem of weighting, or adding together response.
This problem is caused by an unequal number of responses per subject,
different degrees of certainty about succeeding responses, etc.

Each of these methods has its drawbacks. Deese's method allows
only for comparison between groups. Noble's method encourages only
chaining--that is, giving an association to the prior associate rather

than to the original stimulus word.3

Also there remain Deese's objections
about combining a dominant initial response with an idiosyncratic late
response, and weighting unequal numbers of responses from subjects.
Garskof and Houston (1963) describe a method designed to eliminate
all of these objections to Noble's method. Chaining is discouraged by

having subjects write each of their associations immediately after the

 

5As long as it is constant for all words, the occurrence of chaining can
often be ignored. For this study, however, we desired to get as pure a
measure as possible of the meaning of Rokeach's values.

-14-
stimulus word, which is repeated several times on the same page.
Whenever responses from different subjects are compared, they are weighted
by their order of iemission, thus accounting for unequal numbers of
responses and for intensity of response. This is in accordance with the
extension of Marbe's law to the area of verbal association by Bousfield
_e_§ LL; (1956) .

Garskof and Houston's method of data collection was employed in
pilot work for this study. Although the determination of chaining was
left to the intuition of the experimenter, it seemed evident from series
of responses such as "Marines, air force, men general, private" (which
were given to the "National Security") that chaining had occurred despite
Garskof and Houston's precaution against it.

Th2,Similarity g£_Associative Hierarchies

Although Noble and Deese were interested in meaningfulness primarily
as a variable in verbal learning studies, it soon became evident that
their techniques provided material by which one could determine whether
two words (or the same word to two people) had the same meaning. Several
measures now exist for assessing the similarity of two or more response
hierarchies. The most complex is that of Garskof and Houston (1963)
which-~as already mentioned--takes into account the order of the elements
(or words) in the associative response hierarchies and allows for any
number of associations. Not including this measure, Marshall and Cofer
(1963) list ten measures for calculating the degree of similarity in
response hierarchies. These differ among themselves mainly in such
variables as group versus individual data, one or more responses per
subject, etc. All of them give an index of the similarity between two or

more response hierarchies.

Although.the method of obtaining associations to words allows one

-15-
to quantify the degree of similarity that concepts have, in general

this technique results in an overflow of different associations to the

same concept. One alternative is to group the several responses on the
basis of some outside category to which they belong (e.g., via a thesaurus),
but this may logically defeat one's purpose by calling connotatively
different responses equivalent. The result is that the association
technique only shows the degree to which two concepts are or are not

alike, except by simply reporting the responses that were obtained.

A technique for measuring meaning which goes a long way toward
overcoming this particular limitation of associative response hierarchies
is the semantic differential (Osgood, Suci, and Tannenbaum, 1957). Rather
than produce associations to the concept being measured, subjects are
presented with pairs of associations. These pairs are presumed to be
opposites (good-bad, high-low, strongdweak), and subjects are asked to
judge the degree and direction of the relationship between the pairs of
associations and the concept being measured (see Appendix C). The
semantic differential allows one to compare the associative structure of
words with respect to a manageable number of concepts. The semantic
differential originated within the framework of a specific theory of
'meaning (meaning is a representational mediation process) which is
somewhat different in focus from the thought behind free association
techniques.

However, for use as a technique neither the semantic differential
nor free association need say anything about a theory of association as
such. This, in fact, is pointed out by both Deese (1965) and Osgood 25

g. (1957) .

-16-

The semantic differential is a technique rather than a specific
measuring instrument. It requires an optional number (from 3 to 25 or
more) of seven-point bi-polar scales (such as good-bad, hot-cold, up-down)
on which the subjects may judge concepts, nonsense syllables, pictures,
etc. (see Appendix C for a typical set of instructions). In the original
research with the semantic differential and in most of the subsequent work
it was found that when a large number of scales were used, three factors
(labeled evaluation, potency, and activity) accounted for most of the
variance.

This simplifies measuring the meaning of concepts, since the meaning
can be described in terms of these three factors. Of course such quanti-
fication by no means exhausts the meaning of a concept. As Osgood points'
out, the semantic differential is concerned with what is usually referred
to as connotative rather than denotative meaning. This results in such
apparent anomolies as "hero" and "success" having the same meaning when
looked at in the three-dimensional semantic space which the three factors
define. Consequently this puts a limdt on the confidence with which we
can say that two concepts have the same meaning as a result of their
semantic-differential profiles. But it does not limit our ability to say
that two concepts have different connotative meanings if their profiles
are in fact different; and furthermore, in this case (as opposed to use
of associative response hierarchies) we have an operational measure for
specifying both how much and in what respects they differ.

Application 5;; Measuring Techniques 53 1.9.1225.
When specific uses of the semantic differential are concerned, an

additional problem.may arise. Osgood g£_§l, (1957) speak of "concept-scale

-17-
interaction" which may distort the factor structure of a set of scales.
For example, it was found that when the sample of concepts to be judged
was limited to aesthetic objects (paintings), a type of activity factor
becomes very dominant. Also, when judgments were limited to sociopolitical
concepts the activity and potency factors seemed to join in a dynamism
factor. The dynamism factor was also found to be evident in a study by
Osgood, Ware, and Morris (1961) in which each of Morris' 13 "ways to
live" were judged against 26 semantic differential scales. All three
of the traditional factors (evaluation, potency, and activity) fused into
a successfulness, or dynamism factor which was the principal factor. A
product-moment correlation of .66 was found between the dynamism factor
and individual value preference (as measured by the "ways to live").
There was no correlation with the other two factors that were obtained
(which were labeled "predictability" and "kindness"). The authors hypoth-
esized that the reason for the results showing one large dynamdsm.factor
was that in this culture there was a preference for a life pattern that
was strong and active.

It should also be remembered that Morris' "ways to live" are by
no means unitary concepts, but include goals as well as means (Morris and
Jones, 1955). Osgood, Ware and Morris were not concerned with whether
subjects agreed among themselves on the meaning of the "ways to live."
Indeed that particular problem is not likely to arise with the "ways
to live" since they are long descriptive statements.

Kelly (1955), however, points out that agreement on the meaning of
words may be a very great problem, especially when these words are personal
constructs-~trait names used by the individual to describe himself.

Loehlin (1961) attempted to find out just how much such trait

-18-
names did differ in their meaning from subject to subject. Although

not treated as "values" in this orientation, the trait names are
sufficiently close to Rokeach's instrumental values to make Loehlin's

work very relevant. Trait names were taken from Cattell's (1957)
"personality sphere" and included such concepts as: complaining, dominant,
imaginative, relaxed, inconsiderate, and expressive. Loehlin measured
subjects' agreement of the meaning of these concepts using Osgood's
semantic differential as well as two other measures of concept similarity
("direct rating of similarities among adjective meanings" and "rating of
other persons well known to the subject, using the adjectives [Ehe concepts
or traits:7, and comparison of the self to these persons").

Loehlin found that subjects differed among themselves in the meaning
of these concepts just as much as they differed in assigning the concepts
(traits) to themselves. Then Loehlin (1967) repeated the study using
triads of synonymous adjectives instead of single adjectives. For example,
he replaced "relaxed" with "relaxed, easy-going, placid." The addition
of synonyms did not increase subjects' agreement on the meaning of the
words, and the findings of the earlier study were replicated. Loehlin
concluded, "A considerable improvement (in the agreement on the meaning
of the concepts being used) would be necessary before differential
inferences about self-concepts could be confidently made."

If an analogous situation should hold for Rokeach's value scales it
would mean that subjects differed among themselves in the meaning of
a value like "imaginative" just as much as they differed in their ranking

of the value "imaginative."

-19-
pig 9: "Meaning" is; This Study

For the purpose of this study we will define meaning as either
connotative meaning or as associative meaning.4 The connotative
meaning of a word is operationally defined as that which is measured by
the semantic differential, and the associative meaning of a word is
operationally defined as the verbal responses which subjects give to
that word.

Since associative meaning may refer to all of the verbal responses
which a subject gives to a word, this includes both what is commonly
referred to as connotative and also denotative meaning. For this study,
however, it was desirable to restrict associative meaning insofar as
possible to the objective referents of the values. Accordingly, the
typical instructions for obtaining associative response hierarchies were
modified. Rather than ask for free associations, the subjects were
asked to give words that would tell what the concepts (values) meant
(see page 26). It was hoped that these instructions would eliminate
such purely evaluative words as "good," "nice," "bad," "awful," etc.,
which tell nothing about the denotative meaning of the concepts. Whether
these instructions do, in fact, result in a different quality of responses

than free associations has not yet been determined.

Purposes _§,This Study

 

The purpose of this thesis is to investigate the connotative and
associative meaning of the values on Rokeach's terminal and instrumental

value scales. There are four specific questions which we hope to answer:

 

"We sidestep the issue as to whether these two types of meaning involve
different processes. See Bousfield (1961) and Osgood (1961) for the
controversy over this point. When we simply use the term "meaning" it
should be taken to include both connotative and associative meaning.

-20-

1) Does the rank ordering of the values accurately reflect
the subjects' evaluation of these concepts, as measured
by the semantic differential?

2) What is the relation of the value ranking to other
(non-evaluative) dimensions of the semantic differential?

3) Are there any systematic differences in the associative
meaning of a particular value for a group of individuals
if some rank that value high and others rank that value
low?

4) To what extent can we say that the values mean the same
thing--associatively and connotatively-~to different people
(regardless of how they rank them)?

The purposes of this study have been expressed in the form of
questions rather than specific hypotheses because the problems are not
independent. That is, the answer that we give to one question may affect
the way in which we attempt to answer the remaining questions. The inter-
relationships among the questions are explained below.

The first two questions refer to connotative meaning. They concern
the validity of the ranking procedure for measuring values, and use
only the semantic differential. If Rokeach's value scales are valid
measures of a person's value preferences, then we expect that an individual
subject's value ratings on the evaluative scales of the semantic differen-
tial will correlate highly with his value preferences as indicated on the
value scale. Furthermore, we would not expect to find a strong relation-
ship between the value scales and the other two semantic differential fac-
tors .

Each of the first two questions, then, is concerned with whether the

-21-

connotative meaning of the values is changing with the rank of the
values. The third question is also focused on the ranks of the values,
but it is concerned with shifts in associative (rather than connotative)
meaning. Thus a value may have different referents for a person who
ranks that value high than it does for a person who ranks that value low.

The fourth question is concerned with a different way in which the
meaning of the values might differ. Namely, two people could rank a
value the same, but still give it different meanings. If this possibility
should be the case, it is equivalent to saying that no matter where a
person ranks a value one cannot be sure what the value means to him.
Therefore, the fourth question asks whether the values mean the same
(connotatively and associatively) to everyone.

We have already pointed out that meaning must be measured indirectly--
by means of associations, the semantic differential, or some such technique.
Thus when we answer the questions, "Do the values mean the same to every-
one?" we must answer it in terms of the techniques that we employ. For
the associative response hierarchies5 (associative meaning) this means
reporting some index of similarity that is based on equivalencies of
obtained responses. For the semantic differential (connotative meaning)
this means this means looking at the variance of the mean scores for each
of the factors being measured. But it is important to note that in both
cases our index of similarity will be uninterpretable unless it can be

referred to some baseline.

 

'SThis term "associative response hierarchy," properly refers to single

word verbal associates to a particular concept that are obtained by free
association. We retain the use of the term here even though our technique
is not free association.

-22-

In other words, we cannot directly answer the question, "Do these
values mean the same to everyone?" Rather we must re-phrase it to
read, "Do these values mean the same to everyone, at least to the extent
that common 'everyday' words do?" The implicit assumption behind this
approach to the question is that most words convey about the same meaning
to everyone-~at least to the extent that verbal communication is sufficiently
accurate for most purposes.

Consequently it was necessary to obtain a set of "control words"
that could provide a baseline for the similarity measures. In order to
make the comparison as fair as possible, control words were equated with
the values for frequency of usage and for meaningfulness (2'). The
procedure for doing this is contained in the method section below.

Our fourth question, then, will be answered by comparing data from
the values with data from the control words. This is in distinction
to the first three questions which will be answered only by looking at
data from the values. These questions are not entirely independent,
however. If it is found in answering any of the first three questions
that meaning differs as a function of rank, this indicates that the
group as a whole does not agree on the meaning of the value.

But if we fail to find any relationship between the rank or importance
and the meaning of a value, we still do not know anything about the extent
to which the group as a whole agrees on the meaning of that value; the dif-
ferent subjects could be anywhere from perfect agreement to complete dis-
agreement on the meaning of the value. In any case there could be no

relationship between the rank of the values and the meaning of the values.

-23-
METHOD

Subjects

The study consisted of eight parts, labeled A, B, C, D, E, F,
G, and H. Two distinct groups of subjects were used for these parts.
The first group consisted of 176 subjects enrolled in a social psychology
course at Michigan State University. This group was randomly divided
into four‘subgroups of 44 subjects each; each subgroup was used for one
of the first four parts of the study (A, B, C, or D).

The second group of subjects consisted of 88 students enrolled in
an introductory psychology course at Michigan State University. These
subjects were randomly divided into four groups of 22 subjects each; each
group was used for one of the last four parts of the experiment (E, F,
G, or H).

Initial Procedure

For the first four parts of the study, four sets of test booklets
were passed out to the subjects. There was a different set of test
booklets for each part of the experiment: A, B, C, or D. Each set
of test {booklets was given to 44 different subjects. The first page of
each booklet consisted of one of Rokeach's two value scales. The subjects
for parts B and D were given the instrumental value scale. The following
instructions for filling out the value scale were on the first page of the
test booklets:

Below is a list of 18 values arranged in alphabetical
order. We are interested in finding out how important
each of these values is for you, as a guiding principle

in your life.

First, study the whole list carefully. Then pick out
the six values which are most important to you, and check

'5

-24-
each of them in column 1.

Next, pick out the six values which are least important
to you, and check each of them in column 3.

Finally, check the six remaining values in column 2.

If you céhhge your mind, feel free to change your answers.
Please take all the time you need to think about your
answers, so that the choices you make truly reflect the way
you feel about these values.

Immediately following these instructions came a list of 18 values
(either the terminal or the instrumental values-~see Appendix A) with
three columns of boxes to the left of the values where subjects could
make their checks. Below the list of values were the following instruc-
tions:

Now study carefully the six values you have checked in
column 1. Place a'l next to the value which you feel

is the most important of these six values. Place a'g

next to the value which is second most important to you.
Continue in this way until you have ranked the six values
in column 1 from‘l to 6.

Then go to column 2 and rank the six middle values from‘Z
to 12, Then go to column 3 and rank the six least important
values fromll; tollfi.

When you have finished, go back and check over your rank-
ings. Please make sure that every value has a ranking, and
that no number has been left out or used twice.

After they had completed the value scale, subjects continued to
fill out the test booklets. Two sets of booklets (for parts A and B)
were for obtaining associations to each of the 36 values: one set for
the 18 terminal values, the other set for the 18 instrumental values.
Two other sets of booklets (for parts C and D) were used to obtain
semantic differential ratings for each of the 36 values; again, one

set was for the terminal values and one set for the instrumental values.

Part A; Associations §2_Terminal Values

 

 

Since pilot data indicated that a pure free association technique

-25-
tended to produce both chaining of responses and simple evaluative
words (e.g., "good," "bad," senseless," etc.) which had little to
do with the denotative meaning of the values to the subjects, the typical
association technique was modified so that subjects would give responses
which defined the value for them. Single-word responses were, however,
encouraged. The instructions were:
Below is a list of 18 values. Beneath each value there are
five blanks. Please fill in these blanks with words or
phrases tlat show what each of these values mean.
Whenever possible use single words--such as common associ-
ations or words that symbolize the value. Do not use the
words already contained in the value itself or in the
value's explanatory phrase given in parentheses. If it is
necessary to use more than one word for any blank, please
be as brief as possible.
Do not leave any of the blanks empty. Give five brief
associations for each value which will tell what the value
means.
The 18 terminal values followed these instructions, with five blanks
below each value where subjects could write their responses. For

example:

A COMFORTABLE LIFE (a prosperous life)

 

AN EXCITING LIFE (a stimulating, active life)

 

ETC.
Subjects were given as long as they needed to finish this part.
The average time was about 30 minutes.
Egg; _B_: Associations 1:2 Instrumental y_a_1_];t_1_e_s_

This part was exactly the same as part A above, except that the

-25-
instrumental values replaced the terminal values.
Egg; 9: Semantic. Diffejrfential Rati_r_1g§_ _f_c_>_r_ _t_l_1_e_ Terminal Values

Since it was desirable to measure the values on the three factors
which are traditionally found with the semantic differential, an 18-
scale semantic differential was constructed. The 18 scales were selected
from.previous work by Osgood g_“gl. (1957 and by Osgood ggugl. (1961).
Six scales were selected to measure each of the three factors (evaluation,
potency, and activity). Criteria for scale selection were that the
scale load highly on the factor which it was intended to measure, and
that the scale not load highly on the other factors. This large a
number of scales was chosen (three scales per factor is more usual) in
order to insure that differences between values could be obtained for
each subject, and also because concept-scale interaction might invalidate
some of the scales.

The scales selected to measure evaluation were: beautiful-ugly,
timely-untimely, kind-cruel, good-bad, positive-negative, and successful-
unsuccessful. The scales for potency were: heavy-light, hard-soft, large-
small, strong-weak, masculine-feminine, and powerful-powerless. The
scales for activity were: excitable-calm, fast-slow, hot-cold, active-
passive, vibrant-still, and warmecook.

Half of the scales for each factor were reversed (e.g., "bad-good"
instead of "good-bad"), and the order of the scales was randomized. A
test booklet was constructed in which each of the terminal values was
paired with each of the scales. A separate page was used for each value.

Thus the first page would look like this:

-27-

A COMFORTABLE LIFE (a prosperous life)

 

heavy : : : : : : light
soft : : : : : : hard
beautiful : : : : : : ugly

 

 

excitable etc.
All 18 scales were listed below each value. Subjects were instructed
to rate each value by simply placing a check on each of the 18 scales.
The instructions, adapted from Osgood ggugl. (1957) are given in
Appendix C.
.2335.2: Semantic Differential Ratings £93,5hg.lnstrumental Values

The procedure for this section was exactly the same as for part C
above, except that the instrumental values were substituted for the
temminal values.

Obtaining Control Words

 

As we mentioned earlier, it would be necessary to obtain similar
data for a set of "control" words. To obtain these control words,

75 words were randomly selected from.the Thorndike-Lorge (1944) word
count, with the restriction that the words be either nouns or adjectives,
and also be of the same frequency-of-usage range as the values.

At this point an additional problem had to be dealt with. Besides
varying in clarity (or consistency of meaning), words also vary in their
meaningfulness. As we mentioned earlier, this is the major variable
with which much of Noble's work has been concerned. And, since a large
Q value (or a greater number of associations) would be equivalent to a
more heterogeneous associative response hierarchy, this could cause a

word to seemambiguous when different subjects' response hierarchies are

-23-
compared. Therefore, after the control words had been selected, they
had to be equated with the values with respect to the number of associa-
tions they were capable of eliciting.
In order to do this, the 75 control words together with the 36

values were measured on their meaningfulness, using Noble's‘g'.
Twenty subjects (taken from an introductory psychology course at
Michigan State University) were used to obtain the‘gf measure on each of
the control words and the values. The instructions, adapted from
Noble §£_§l, (1957) were as follows:

This is a test to find out how many associations a person

has to certain words or phrases. Below is a 125 item list;

you are to rate each item as to the number of things or

ideas it makes you think of. The ratings are made by

simply placing a number from 1 to 9 in the blank before each

item. Use the following scale for a guideline:

: 1 : 2 : 3 : 4 : 5 : 6 : 7 : 8 : 9 :

 

very many an average very few
associations number of associations
associations

Look over the list quickly to get an idea of the type of
words and phrases which it contains, then simply rate each
of the items as to the number of things or ideas that it
calls to mind. For instance, compare the words "food" and
"delimit." The former probably calls many things to mind,
and you would put a.1 or a g_in front of it. The word "de-
limit," however, probably calls few things to mind, and you
would put an‘g or a 2_in front of it. Try not to let any

of your own feelings about the appropriateness of the concept
influence your ratings. Be as accurate in your ratings as
you can, for the results will be used for research purposes.
We are interested in ratings given by people in general, and
not in your own personal scores. This is not a speed test.
You will be given plenty of time to finish. Any questions?

The 111 items (75 control words and 36 values) were preceded by
14 items of the same kind which served the purpose of letting the sub-

jects become familiarized with the task.

-29-
After the mean.m' values were obtained, 15 words were selected
at random from those control words that were within the same.g' range
as the values. These 15 words were then used as control or baseline
words against which to compare the values on the degree of agreement
of the associative response hierarchies and the semantic differential
ratings. The 15 words, in the order in which they were selected and
used, were: fat, refrigerator, death, brother, thunderstorm, utensil,
month, bride, smooth, tongue, apple, flag, orgy, brown, boy.
§a_r_§ E: Baseline _D_§_t_:_a_ £9; Associations
Twenty-two subjects were given the list of control words and

asked to give associations that would define the words, just as the
subjects in parts A and B had been asked to define the values. The
instructions were modified slightly, as given below:

Below is a list of 18 words or concepts. Beneath each of

them there are five blanks. Please fill in these blanks

with words or phrases that show what each of these concepts

mean.

Whenever possible use single words--such as common associates

or words that symbolize the concept. Do not use another form

of the word that is already given (like "fatty" for "fat').

If it is necessary to use more than one word for any blank,
please be as brief as possible.

Do not leave any of the blanks empty. For each word or
concept, give five brief associations that will tell what
it means.
The 15 control words followed the instructions, with five blanks
below each word, just as had been done for the values in parts A and B.
-PartԤ: Baseline Data for Semantic Differential Ratings

Another 22 subjects were used for obtaining baseline data on the

semantic differential. The procedure for this part was exactly the

-30-
same as for obtaining semantic differential ratings for the values
(parts C and D above). The exact same instructions were used, except
that the phrase "words or concepts" was substituted for "values" where
appropriate.
Deletion p£_Values' Explanatory Phrases

After baseline data had been obtained on the control words for
both associative response hierarchies and the semantic differential,
an additional problem had to be dealt with. It seemed doubtful that
the control words-~being single words rather than phrases or pairs
of words--were a fair comparison for the values. There were two factors
which could conceivably make the values seem to have less similarity
of denotative meaning than the control words. First of all, each value's
explanatory phrase eliminated one or two responses which might have been
given by most of the subjects. Also, there was the possibility that the
values-~being complex stimuli--might be more difficult to respond to
(even though they could still be more well-defined grammatically). On
the other hand, the concern of this thesis is focused on the values
exactly as they appear on the value scales. Therefore, it was decided
to retain the explanatory phrases for the values in every instance except
for their comparison with the control words on the similarity of their
associative response hierarchies. Parts G and H of the study were added
to obtain the associative response hierarchies of the values when the values
are presented without explanatory phrases. [
fig; Q: Associations j_:_o_ Terminal 1733133 without Explanatory Phrases

A third group of 22 subjects was used to obtain associative

response hierarchies for the terminal values without the explanatory

-31-
phrases. For example, instead of being asked to define "A COMFORTABLE
LIFE (a prosperous life)" subjects were simply asked to define "A
COMFORTABLE LIFE." Instructions were the same as for part A.

A final group of 22 subjects was used to obtain associative response
hierarchies for the instrumental values without the explanatory phrases.
For example, instead of being asked to define "AMBITIOUS (hard-working,
aspiring)" subjects were simply asked to define "AMBITIOUS." Instruc-

tions were the same as for part B.

RESULTS

Mg ph_e Semantic Differential Egg; Structure

In order to check whether concept-scale interaction might have
invalidated results with the semantic differential, a factor analysis
was carried out on the correlation matrix of the 18 semantic differential
scales. The principle concern was that the three factors that had been
built into the 18 scales might not emerge. As recommended by Osgood
,_p‘p_. (1957) the factor analysis was performed on the scales across all
subjects and values taken together. Thus there were 1564 observations
(18 values times 88 subjects) per each variable (or scale). Verimax
rotation was used. With a few exceptions, the three sets of scales did
load on different factors respectively. (See Table 1, page 33.) Note
that the scales are not listed in this table in the order in which they
were presented to the subjects during the experiment. Since three scales

(timely-untimely, powerful-powerless, and warm-cool) either loaded poorly

on their intended factor, or just as highly on another factor, they were

-32-
dropped from further analysis. This left five scales to measure each
of the factors.6

Even after discarding three scales, the remaining scales were, of
course, far from being pure scales (that is, they measure more than one
factor). Therefore a check was made to determine whether the scales for -
potency and activity might also be measuring evaluation (since this was
the most important contamination to avoid). Fischer's r-to-Z transforma-
tion was used to compute the mean correlations between the three sets of
scales and the three factors. These results (see Table 2, page 34) give
an estimate of the degree of dependence between the three factors, as
measured by the five scales for those factors. Thus the five scales
which measure potency correlate .286 with the evaluative factor, and we
can expect our measure of potency-~which should be independent--to corre-
late .286 with the evaluative factor.

Therefore, since the scales measuring both the potency and activity
factors were somewhat "contaminated" with the evaluative factor, a smaller
set of six scales was selected to use whenever that independence was
necessary. (For example, we cannot determine the relationship between the
rank order of the values and potency if our measure of potency is con-
taminated with evaluation.) These scales were "good-bad" and "kind-
cruel" for evaluation; "hard-soft" and "heavy-light" for potency, and
"excitable-calm" and "vibrant-still" for activity. The two evaluative

scales were chosen to match the potency and activity scales in their

 

6' t of obtainin comparable results for all three factors,
1:"wzged::I:::Ie to have thegsame number of scales for each. The decision
to retain large-small rather than powerful-powerless on the potency factor
was made on the basis of further factor solutions, in which large-small
increased its relative loading on potency over evaluation while powerful; d
powerless did the opposite. On the five factor solution large-small Izade
.3775 on potency and .3560 on evaluation, while powerful-powerless oa e
.6306 on evaluation and .3843 on potency.

-33-

Table l. Three-factor verimax solution of the 18 semantic differential
scales used in this study.

 

 

 

 

Factor I Factor 11 Factor III
Scalgs 1,2477)* (.1604)? (£1631)*
good-bad .8101 .0141 .1128
c: positive-negative .7981 .0392 .1891
'43 successful-unsuccessful .7040 .2925 .1825
g beautiful-ugly .6863 -.3980 .1774
“3 kind-cruel .6859 -.4406 .2809
timely-untimely .4335 .1821 .0169
hard-soft -.O647 .7951 .1101
heavy-light .0645 .6880 .0097
g masculine-feminine .1954 .6429 .1131
:2; strong-weak .5781 .4908 .2832
large-small .5451 .2626 .2237
powerful-powerless .6035 .4653 .2617
excitable-calm -.0453 -.0544 .7902
vibrant-still .3275 .1522 .7535
:3 fast-slow --.0117 .3224 .6705
in: active-passive .3111 .3873 .6599
hot-cold .2973 -.O704 .5795
warm-cool .4519 -.2906 .4302

 

*Proportion of variance accounted for.

-34-

Table 2. Mean correlation of the three sets of five scales
with the three factors.

 

Factor
Scales Evaluation Potency Activity
Evaluative 0742 " 0139 0126
Potency .286 .604 .144
ACtiVity 0180 0155 0699

 

 

-35-
correlation with their respective factors. The selection of the potency
and activity scales was made to maximize their independence with evalua-
tion. The information given in Table 2 is repeated in Table 3 on page

36 for this shortened six-scale semantic differential.

Scoring ppg_Associations

Subjects were compared on both the association and the semantic
differential ratings in order to determine the degree of similarity and
hence the degree to which the words (values) meant the same to the
different subjects. The same analysis was carried out for the control
words.

For the associations that the subjects gave to define the values,
the similarity was measured by Jenkins and Cofer's (1957) Measure of
Relatedness (MR). Expressed as a formula this is

MR = Rc/Rt
where Rc represents the number of responses common to two different
subjects, and Rt is the total number of responses that were given by
both subjects. For example, suppose that two subjects gave the following
sets of responses to the value "A comfortable life:"

First subject: money, leisure, rest, security
Second subject: rest, fun, home, family, money

Then the MR for "A comfortable life" for these two subjects would be

4/9, or .444--since four of the nine responses are matched.7

In order for words or phrases to qualify as equivalent, the exact

H "rOCk," "rocky,"

same root word had to be present. For example, "rocks,
and "full of rocks" would all be coded alike, but "big stone" or "boulder"

would not be coded with them. All phrases were coded under what was

 

7Despite instructions, subjects often did not give five responses.

-36-

Table 3. Mean correlation of the three sets of
scales with the three factors.

two "independent"

 

 

Factor
Scales Evaluation Potency Activity
Evaluative .746 -.191. .182
Patency " 0000 0744 .054
ACtiVity -0026 0138 0734

 

-37-
judged to be the dominant word in the phrase. Negative expressions of
a word were kept separate from positive expressions, and all means of
expressing negation were coded as equivalent. Thus "breathless," "no
breath" and "out of breath" would all be coded as equivalent.

The mean MR's for the 44 subjects were computed on all 36 values.
The mean MR may be looked at as follows. Since there are 44 subjects,
each subject will have an MR with the other 43 subjects. This gives a
total of 946 different MR's. The mean MR is simply the sum of all these
MR's divided by 946. These results are given in Tables 4 and 5 (pages
38 and 39) for each of the terminal and instrumental values respectively.8

In order to determine whether or not there were any systematic
differences in the agreement on the denotative meaning of the particular
values between different subjects as a function of the importance (or
rank) given the values, the mean MR was also computed for the upper and
lower quartile of the distribution of ranks that had been given to each
value. Also the mean MR that resulted from comparing the upper quartile
with the lower quartile (referred to as the between-quartile mean MR)
was computed. (For a clarification of these quartiles, see Figure 1,
page .) These results are also given in Tables 4 and 5 (pages 38 and 39)
for the terminal and instrumental values respectively.

That is, for each value, the 44 subjects were ordered from.highest
to lowest according to the rank that they had assigned to the value. Then
the mean MR was calculated for the upper quartile (11 subjects, or 45

different MR's), the lower quartile (also 11 subjects or 45 different MR's),

 

8Note that the mean MR for the total groups is given here only for the
sake of completeness. When the total group's mean MR is compared with
the control words, the data in Tables 17 & 18 (pp. , ) will be used.

-33-

Table 4. Mean MR (measure of relatedness) for terminal values, for total
group, upper quartile, between quartile, and lower quartile.

 

Between L Lower
Quartile Quartile
(N = 121) (N - 55)

Total Upper
Value Group Quartile
(N* 8 946) (N = 55)

 

 

A comfortable life .0878 .0999 . .0722 . .0545
An exciting life .0607 .0645 .0553 .0291
A sense of accomplishment .0618 .0460 .0634 .0582
A world at peace .0588 .0747 .0653 .0573
A world of beauty .0470 .0764 .0418 .0117
Equality .0748 .0488 .0664 .0732
Family security .0826 .1088 .1062 .0732
Freedom .0301 .1394 .0380 .0251
Happiness .1084 .1395 .1013 .0542
Inner harmony .0553 .0816 .0537 .0533
Mature love .0536 .0594 .0269 .0284
National security .0412 .0397 .0309 .0498
Pleasure .0635 .1431 .0751 .0299
Salvation .0727 .1175 .0566 .0218
Social recognition .0470 .0239 .0292 .0230
Self-respect .0495 .0408 .0547 .0389
True friendship .0655 .0534 .0432 .0606
Wisdom .0842 .0549 .0613 .0741
Mean .06358 .07845 .05786 .04535

 

* "N" refers to the number of individual MR's that determine each column
entry; the number of subjects involved is 44, 11, 22, and 11 respectively.

-39-

Table 5. Mean MR (measure of relatedness) for instrumental values,

for total group, upper quartile, between quartile, and

lower quartile.

 

 

 

Total Upper Between Lower

Value Group Quartile Quartile Quartile

(N* = 946) (N = 55) (N - 55) (N = 55)
Ambitious .0463 .0327 .0324 .0340
Broadminded .0916 .0721 .0792 .1351
Capable .0746 .0860 .0594 .0637
Cheerful .1602 .2525 .1637 .1123
Clean .0563 .0530 .0516 .0506
Courageous .0626 .1149 .1157 .0619
Forgiving .0958 .0845 .0507 .0272
Helpful .0612 .0722 .0467 .0271
Honest .0504 .0338 .0499 .0382
Imaginative .0506 .0238 .0399 .0404
Independent .0476 .0564 .0426 .0294
Intellectual .0649 .0400 .0505 .0847
Logical .0200 .0149 .0061 .0036
Loving .0631 .0909 .0673 .0289
Obedient .0271 .0462 .0210 .0249
Polite .0500 .0563 .0610 .0544
Responsible .0818 .1207 .0685 .0462
Self-controlled .0243 .0136 .0168 .0172
Mean .06268 .07025 .05663 .04887

 

*"N" refers to the number of individual MR's that determine each column
entry; the number of subjects involved is 44, ll, 22, and 11 respectively.

-40-

Figure 1. Schematic diagram of the distribution of MR's for any partic-

Symbols:

X:

B:

ular value.

LSHigh ranks) - - - - X - - - - (Low ranks)

 

B D

 

 

 

  
    

44 ranks which the subjects have
given to this particular value,
arranged in descending order.

this section (which includes sections
B, C, and D) contains 946 individual MR's,
representing all possible comparisons
between the 44 subjects.

 

This section includes 55 individual MR's, computed
only on the upper quartile of the distribution of
ranks (the number 55 represents all possible com-
parisons between 11 subjects).

this section is exactly the same as section B, ex-
cept that it is taken from.the lower quartile of
the distribution of ranks.

this section includes 121 individual MR's, representing every
possible comparison between those subjects who constitute the
lower quartile with those subjects who constitute the upper
quartile; the mean of this distribution is the mean between,
quartile MR.

(saws: m1) - - - - x - - - - (anus: 113m)

 

I‘I

-41-

and for the cross-section of these quartiles (11 subjects by 11 subjects,
or 121 different MR's).

The MR for the cross-section of the upper and lower quartiles will
be referred to as the between-quartile MR. The relationship of the between-.
quartile MR to the upper and lower quartile MR's is all important for
determining whether the associatiVe meaning of values varies systematically
with rank (question 3). For example, suppose that the between-quartile
MR should be significantly lower than the upper and lower quartile MR's.
This means that the responses which a subject gives to a value are relatively
more like the responses given by other subjects who have ranked the value
similarly to him. And, conversely, the responses are relatively less
like the responses given by subjects who have ranked the value differently.
Another way of putting this would be to say that subjects in the upper
quartile tend to agree with other subjects in the upper quartile, and
subjects in the lower quartile tend to agree with other subjects in the
lower quartile.
Computations with.£hg.3emantic Differential

0n the semantic differential data, each value was scored first
of all for each subject, and given a score on evaluation, potency,
and activity. The scores for each factor were the sum of the five
scales used to measure that factor. For each factor this gives a possible
range of 5 to 35, with 20 representing neutrality (each individual scale
was scored from 1 to 7, with the low numbers representing the positive
pole of the three factors).

Then the mean score for the total group was computed for three factors
on each value. The variances of these scores tell us how much the subjects

disagreed on the connotative meaning of the value. Thus a high variance is

-42-
analogous to a low MR. (The three variances give three separate indices
of the degree of disagreement, each one related to a particular factor or
aspect of connotative meaning.) Just as for the MR measure, the mean and
variance of the three semantic differential factors were also computed for
the upper and the lower quartile of the distribution of ranks for each
value. Since individual scores are not obtained by comparison with other
subjects for the semantic differential (as they are for the associations)
there is no between-quartile measure in this case. The semantic differen-
tial data are shown in Tables 6 and 7 on pages 43 and 44.for each of the
terminal and instrumental values respectively.
Normative gay; 93 the Associative Response Hierarchies

Although the actual associations given to each of the values in
parts A and B were not relevant to this study, it is likely that they may
be of use in further studies which center on a particular value. Therefore
the ten most frequent associations to each value, as well as the breakdown
of these associations by upper and lower quartile, are given in Appendix D.
Results Specific to Question‘l

We intend to answer the four questions separately for the terminal
and instrumental values. Since there is a theoretical distinction between
the two types of values, there is no reason to presuppose that a question
cannot be answered one way for the terminal values and another way for
the instrumental values. The first question is, "Does the rank ordering
of the values accurately reflect the subjects' evaluation of these concepts
as measured by the semantic differential?"

The answer to this question can be found by looking at the correlations

between the ranks of the values for each individual and his rating of the

-43-
Table 6. Mean* and variance of terminal values on three semantic
differential factors for total group, upper quartile, and
lower quartile.

 

Total Group Upper Qpartile Lower ngrtile

Mean Variance Mean Variance Mean Variance

 

(N = 44) (N = 11) (N = 11);

Evaluation 13.47 40.30 9.09 10.29 15.63 45.65

A comfortable Potency 21.00 25.76 17.00 17.40 24.00 22.40
life Activity 20.13 43.05 16.18 32.16 22.09 44.29
Evaluation 13.09 13.57 11.63 11.45 13.00 18 .00

An eXCiting POtency 15.61 13 .49 16.09 5049 15054 22067
Evaluation 10.97 19.97 8.00 9.40 13.27 13.21

A sense of Potency 14.00 17.44 11.63 14.65 14.63 20.65
accomplishment Activity 14.61 15.82 12.81 17.76 16.27 14.41
Evaluation 8.59 14.75 8.18 13.56 10.09 16.49

A world at Potency 17.36 27.58 17.63 18.05 17.45 25.67
peace Activity 22.36 38.32 21.18 13.16 22.81 23.36
Evaluation 10.72 16.57 11.00 14.60 11.72 28.21

A world of Potency 22.13 15.14 21.72 8.41 23.18 26.36
beauty Activity 20.15 33.90 20.81 36.36 22.09 22.09
Evaluation 10.15 19.11 8.81 11.36 13.00 20.40

Equality Potency 15.50 17.09 13.81 22.36 16.90 12.49
Activity 17.43 24.01 15.63 25.25 21.00 24.20

Evaluation 11.93 35.36 7.90 6.69 16.18 64.76

Family Potency 18.25 16.28 16.36 12.85 18.72 25.61
security ACti-flty 18.86 23.42 16.36 22045 19.72 2106].
Evaluation 9.43 8.57 7.81 8.76 10.45 10.07

Freedom POtency 14.20 22 .07 14.54 41.47 13.81 20.76
Activity 13.63 21.26 13.72 43.81 13.72 22.81

Evaluation 10.59 20.57 8.27 8.81 12.00 15.20

Happiness POtency 21011 9077 19.36 8.65 22090 8089
Activity 17.75 37.07 13.18 25.76 21.18 30.36

 

Table 6, continued

-44-

 

Total Group Upper Qpartile Lower Quartile

Mean Variance Mean Variance Mean Variance

 

(N - 44)_ (N = 11), (N = 11)
Evaluation 12.18 26.52 11.00 34.00 14.18 24.96
Inner harmony Potency 20.31 13.98 19.36 11.85 20.00 12.80
Activity 21.15 23.76 22.18 26.56 20.18 30.36
Evaluation 9.34 20.32 7.45 5.67 12.54 39.67
Mature love Potency 17.06 18.11 16.90 16.09 17.09 22.69
Activity 15.34 38.41 13.09 41.29 18.18 46.76
Evaluation 16.43 29.59 15.18 26.76 19.09 9.29
National Potency 13.79 34.72 9.90 23.49 18.00 30.00
security Activity 17.63 27.12 14.00 20.60 21.81 22.56
Evaluation 13.90 21.34 11.18 12.56 15.45 13.47
Pleasure Potency 23.15 15.76 20.81 7.36 25.54 19.67
Activity 20.68 41.38 16.36 27.25 23.54 54.47
Evaluation 16.06 55.73 7.27 10.41 20.81 13.36
Salvation Potency 19.15 18.04 16.36 18.05 20.00 10.00
Activity 21.59 32.38 16.18 46.56 21.63 3.05
Evaluation 16.40 27.59 12.27 11.61 18.63 7.05
Social Potency 18.97 18.44 16.45 8.67 20.09 7.09
recognition Activity 17.29 17.88 14.63 7.05 18.63 10.05
I
Evaluation 11.36 18.42 8.45 6.67 15.18 15.76
Self-respect Potency 15.15 14.50 14.18 17.96 17.27 9.81
Evaluation 10.06 14.76 7.27 6.41 12.36 18.65
True Potency 16.54 18.85 14.90 15.89 18.27 15.61
friendship ActiVity 15.47 30.11 13.72 32.01 17.27 27.61
Evaluation 10.02 16.20 7.36 8.25 11.18 22.96
Wisdom Potency 14.43 17.46 13.36 21.65 13.90 15.09
Activity 17.00 36.41 12154 36.27 19.09 50.69

 

*Mean scores represent the sum of 5 scales; the possible range
is from 5 to 35, with low scores representing positive ratings

factor.

A score of 20 represents a neutral judgment.

of scores
on the

-45-

 

 

Table 7. Mean* and variance of instrumental values on three
semantic differential factors for total group, upper
quartile, and lower quartile.
Total Grou U er artile Lower rtile
Mean Variance Mean Variance Mean Variance
Q = 44) m =- 11) (AD In
Evaluation 15.59 11.87 13.72 14.41 19.18 6.36
Ambitious Potency 13.22 11.15 13.72 11.81 12.00 12.80
Activity 12.75 8.70 12.90 3.69 14.72 8.41
Evaluation 11.09 6.17 10.72 7.61 11.45 5.87
Broadminded Potency 17.65 8.92 18.18 9.76 18.00 11.00
Activity 18.54 17.13 16.63 13.65 21.00 24.60
Evaluation 13.11 8.14 11.27 9.81 13.90 9.89
Capable Potency 15.68 10.03 14.72 10.81 16.18 13.96
Activity 16.45 6.67 15.00 8.00- 15.72 3.41
Evaluation 11.88 10.75 9.90 5.09 13.27 7.01
Cheerful Potency 23.34 12.04 22.54 17.07 23.72 7.21
Activity 15.34 11.48 15.45 16.47 14.63 8.25
Evaluation 15.06 16.39 11.72 8.41 16.27 18.81
Clean Potency 21.47 14.02 21.90 7.49 22.81 12.36
Activity 20.47 16.67 19.81 10.16 22.27 16.41
Evaluation 13.38 11.26 12.36 16.85 14.09 10.29
Courageous Potency 13.11 10.84 13.72 8.01 13.81 14.96
Activity 15.43 15.55 16.09 16.89 15.63 24.85
Evaluation 12.47 13.79 11.09 18.29 13.54 13.87
Forgiving Potency 12.34 12.41 20.45 8.07 22.36 7.25
Activity 21.06 12.71 19.81 8.76 20.27 11.21
Evaluation 12.29 11046 11.54 7.07 13.09 17089
Helpful Potency 19.15 13.43 18.00 4.60 20.45 17.27
Activity 16.72 11.78 15.45 8.27 17.54 9.87
Evaluation 11.86 13.60 8.45 3.87 14.90 9.89
Honest Potency 16.95 8.50 15.45 15.87 18.72 3.21
Activity 19.04 14.41 17.63 22.05 20.63 9.85

 

Table 7, continued

-46-

 

Total Group
Mean Variance

Upper Quartile Lower anrtile

Mean Variance Mean Variance

 

(N - 44). (N - 111 (N - 11)
Evaluation 12.18 11.08 11.54 8.27 14.18 11.76
Imaginative Potency 18.86 8.86 19.45 18.47 18.45 6.27
Activity 11.93 12.25 12.09 10.29 11.45 6.67
Evaluation 13.04 9.16 11.36 9.65 13.81 17.36
Independent Potency 14.25 12.05 13.63 16.45 14.72 19.81
Activity 16.56 13.87 16.36 22.25 17.45 18.27
Evaluation 14.09 15.66 11.18 5.56 18.00 8.00
Intellectual Potency 17.09 8.82 16.36 6.45 17.54 11.47
Activity 17.93 23.08 15.72 28.61 20.36 18.05
Evaluation 15.40 17.31 12.45 15.07 18.18 10.96
Logical Potency 15.79 8.30 15.63 6.85 15.27 13.81
Activity 20.47 22.90 17.90 32.49 22.72 24.21
Evaluation 9.93 8.80 7.90 5.29 11.72 7.21
Loving Potency 22.93 16.85 21.36 19.05 22.81 20.76
Activity 16.40 17.03 14.18 12.16 17.27 17.61
Evaluation 20.75 29.68 16.00 14.60 24.00 16.40
Obedient Potency 21.20 14.02 20.09 7.29 21.90 15.29
Activity 24.11 20.70 21.45 19.67 25.09 19.09
Evaluation 15.47 21.27 12.36 8.65 17.90 18.49
Polite Potency 22.20 9.60 23.09 5.89 23.27 16.61
Activity 22.06 15.22 20.18 8.96 24.27 16.21
Evaluation 13.54 12.53 11.45 7.07 15.09 11.89
Responsible Potency 15.11 9.07 13.45 16.67 16.36 5.45
Activity 18.59 11.08 18.18 13.96 19.09 14.29
Evaluation 14.56 19.83 12.54 6.27 16.90 27.09
Self- Potency 15.81 10.47 15.09 9.69 17.54 13.07
controlled Activity 21.84 20.74 19.36 16.05 23.27 13.01

 

*Mean scores represent the sum of 5 scales; the possible range of scores

is from 5 to 35, with low scores representing positive ratings on the

factor.

A score of 20 represents a neutral judgment.

-47-
values on the evaluative factor, as measured by the sum of the five
evaluative scales (see pages 27 and 33). Even the use of five scales
to measure 18 values often resulted in several tied ranks for an indivi-
dual's semantic differential ratings of the values (making a Spearman rho
difficult to interpret). For this reason, and also because correlations
with the shortened-six-scale form of the semantic differential would be
desirable, the Pearson product-moment correlation was used.

The median correlations between the semantic differential and the
terminal and instrumental value scales were .6848 and .6273, respectively
(see Table 8, page 48). These correlations are both significant at
well beyond the .01 level.

Results Specific pg Qpestionug

Our second question is concerned with the relation between the
potency and activity factors and the ranking of the values for individual
subjects. An inspection of Table 8 (page 48) shows that for the 15-sca1e
semantic differential all four median correlations are significant at the
.01 level. However, since the 2 sets of scales used for measuring potency
and activity both correlated with evaluation (see Table 2, page 34), we
cannot be sure that these significant median correlations do not simply
reflect evaluation.

Therefore we may look at these same correlations (see Table 9, page 49)
when they are based on the two scales which were selected to be independent
of evaluation (see page 35, and also Table 3, page 36).

To check whether the use of only two scales per factor was a fair
test of any relationship, the correlations with the evaluative factor

based on two scales were again calculated. The median correlation between

~48-

Table 8. Range and median correlations between terminal and
instrumental value scales and three semantic differential
factors-~15 scale semantic differential.

 

 

 

 

 

1155313
Evaluation Potency, Activity
Lowest r .0632 -.3176 -.3052
Terminal Values Median* r .6848 .3650 .4543
Highest r .9294 .8701 .7840
Lowest r -.1518 -.6051 -.3436
Instrumental Values Median* r .6273 .4689 .3285
Highest r .9014 .8722 .8919

 

*Based on 44 subjects

The .05 and .01 significance levels (one-tailed) for the individual
correlations are .400 and .542 respectively. For the median corre-
lations the .05 and .01 significance levels (one-tailed) are .2493 and
.2584 respectively.

-49-

Table 9. Range and median correlations between terminal and instrumental
value scales and three semantic differential factors--6 scale
semantic differential.

 

ficroap

Evaluation Potency Activity

Lowest r -.0454 -.8412 -.4796
Terminal Values Median* r .6652 .1655 .2496
Highest r .9104 .7769 .7739
Lowest r -.3117 -.6922 -.8297
Instrumental Values Median* r .4714 .2260 .1088
Highest r .8612 .8885 .8753

 

*Based on 44 subjects

The .05 and .01 significance levels (one-tailed) for the individual
correlations are .400 and .542 respectively. For the median corre-
lations the .05 and .01 significance levels (one-tailed) are .2493 and
.2584 respectively.

-50-
evaluation and the terminal values dropped from .6848 to .6652. The
median correlation between evaluation and the instrumental values dropped
from .6273 to .4714. Both median correlations were still significant

at well beyond the .01 level, and we may assume that the use of only

two scales is sufficient.

With the smaller set of semantic differential factors all of the
median correlations with potency and activity are still positive. This
time, however, only one of them (for the activity factor with the tenlinal
values) reaches significance at the .05 level.

Results Specific 1:3 @estion 3

Our third question asks whether there are any systematic differences
in the associative meaning of the particular values for a group of indi-
viduals if some rank that value high and others rank that value low. Or,
in other words, it asks whether the values have the same referents for
those subjects who rank them high as they do for those who rank them low,
and vice versa.

Two things are implied for the associative response hierarchies if
there are systematic differences in the associative meaning of the values
as a function of rank. First of all, the MR's should be relatively high
when computed for those subjects who rank a given value alike. And,
conversely, the MR's should be relative low for subjects who rank a given
value differently.

The strongest test to see if this relationship exists is to compare
the mean MR's for the upper quartile, between quartile, and lower quartile

of the distribution of MR's for each value (see Figure 1, page 40). If,

-51-
as a general rule, systematic differences do exist for the values it
should be reflected in a curvilinear trend (specifically a quadratic
trend with the mean between-quartile MR being lower than the other two
MR's) between the overall means for the three divisions.

A two-way analysis of variance with a test for trend on the mean
MR's for the three divisions was performed for both terminal and instru-
mental values (see Tables 10 and 11, pages 52 and 53 respectively). A
significant trend does exist for both the terminal and instrumental values
(p = .01 and .05 respectively), but it is a linear, not a quadratic trend.
The mean MR taken between quartiles falls between the upper and lower
quartile mean MR's. From this we may only conclude that the associative
response hierarchy for the lower quartile differs from the upper quar-
tile's associative response hierarchy for a given value by being more
heterogeneous. This conclusion is based on the following results (which
are true for both terminal and instrumental values-~see Tables 4 and 5,
pages 38 and 39): the mean MR for the lower quartile is generally lower
than the mean MR for the upper quartile; the mean MR taken between these
two quartiles is not lower than the mean MR for the upper and lower
quartile (as it would have to be if the meaning was different for the two
quartiles) but rather it falls between the two quartiles' mean MR's.

waever, to say that the meaning of the values is more heterogeneous
for the lower quartile is not to explain the nature of the heterogeneity.
We will return to this problem in our discussion of these results.
Results Specific pp Question 4

In answering the first three question we found (1) that the

evaluation of the values differed as a function of rank; (2) that except

-52-

Table 10. Summary table of the analysis of variance testing for the
trend of the mean MR for the terminal values.

 

 

SOURCE ss* df MS F p
Rows (values) 2,026,844 17 119,226 2.163 .05
Columns (quartiles) 1,006,340 2

Linear 986,049 1 986,049 17.891 .01

Quadratic 19,683 1 19,683 .356 n.s.
Error 1,873,854 34 55,113
Totals 4,907,038 53

 

*All entries in Table 4 were multiplied by 1,000.

-53-

Table 11. Summary table of the analysis of variance testing for the
trend of the'mean MR for instrumental values.

 

 

SOURCE 83* (if MS I? p
Rows (values) 7,272,398 I 17 427,788 7.044 .01
Columns (quartiles 420,042 2

Linear 410,164 1 410,164 6.754 .05

Quadratic 10,302 1 10,302 .171 n.s.
Error 2,064,780 34 60,729
Totals 9,757,220 53

 

*All entries in Table 5 were multiplied by 1,000.

 

U
-54-

for activity with the terminal values there was no significant relation-
ship between the two non-evaluative factors of the semantic differential

and the rank of the values; (3) there was no significant difference in

the similarity of the associative meaning of the values as a function of
rank. As we pointed out earlier, these results have no direct bearing on
question 4, which asks whether the values mean the same to different

people regardless of rank. Remember that we are speaking of "same" in

a relative sense; we want to know if values have more or less similarity

of meaning than other, non-value words of the same frequency and meaningful-
ness Qp').

We are able to answer this question by referring both to the semantic
differential and to the associative response hierarchies. Each technique
will provide an index of similarity for the individual values which may
be compared to that same index for the control words. First of all, we
will present the results for connotative meaning, as measured by the
semantic differential.

For the semantic differential the index of similarity is the variance
(or the standard deviation) of the scores on the three factors for any
given value or control word. The mean and variance for the terminal and
instrumental values have been reported in Tables 6 and 7 on pages 43 and
45. The mean and variance for the control words are given in Table 12
on page 55. Comparisons between the values and the control words on all
three semantic differential factors were carried out by.p tests. These
results are summarized in Table 16 on page .

For associative meaning the index of similarity is the mean MR

of the associative response hierarchy. As we mentioned earlier, in this

-55-

Table 12. Mean and variance of control words (given in order of

presentation) on three semantic differential factors.

 

FACTOR

Control Word*

Evaluation

Mean Variance .

Potency
Mean Variance

Activity
Mean Variance

25.61

Fat 21.13 47.83 16.59 15.30 21.22

Refrigerator 15.59 23.96 12.45 21.30 22.81 17.77
Death 19.63 39.86 15.40 26.25 22.54 24.73
Brother 14.77 48.18 16.22 14.18 14.90 21.51
Thunderstorm 18.00 29.61 10.27 19.35 14.04 18.71
Utensil 17.00 17.71 18.95 16.33 20.90 18.94
Mbnth 16.90 15.22 20.50 13.02 18.86 13.83
Bride 9.54 43.68 28.31 27.17 13.36 14.24
Smooth 16.59 12.25 23.36 31.48 22.59 30.91
Tongue 17.40 35.49 23.50 10.26 16.72 16.30
Apple 15.04 15.37 18.95 24.04 23.09 23.22
Flag 13.31 29.75 18.09 26.46 17.27 23.35
Orgy 22.45 58.25 16.95 17.85 12.50 36.26
Brown 19.31 31.75 17.95 9.47 20.09 29.89
Boy 14.31 41.37 15.22 32.66 14.09 15.80

 

*N (number of subjects in control group) = 22.

-56-

case associations were given to the values without their explanatory
phrases. The mean MR's for the values without explanatory phrases are
given in Tables 13 and 14 on pages 57 and 58. The mean MR's for the
control words are given in Table 15 on page 59. Two thests were carried
out comparing the mean MR's of the terminal and instrumental values with
the mean MR's of the control words. These results are also summarized in
Table 16 on page 60. This table shows that the results differ for the
instrumental and terminal values.

For the instrumental values, there is greater agreement in both
connotative and associative meaning than for the control words. This
difference is highly significant with the semantic differential data
(connotative meaning) and not significant for the associative response
hierarchies (associative meaning).

For the terminal values, however, the results are ambiguous. There
is significantly less agreement on the associative meaning of the terminal
values compared to the control words. With the semantic differential,
however, the only significant difference between the control words and
the terminal values is in the evaluative factor-~with the values showing
greater similarity of meaning. The p_ratio for the activity factor just
misses the 5% level of significance (with control words showing greater

similarity of meaning).

DISCUSSION

@estions 1 and _2_

Our answers to the first two questions have been rather straight-

forward. We have found that the ranking procedure--applied to Rokeach's

Table 13. Mean MR (measure of relatedness) for terminal values, without

defining phrases.

-57-

 

 

 

Values (N* = 231)
A comfortable life .1489
An exciting life .0705
A sense of accomplishment .0839
A world at peace .0879
A world of beauty .0825
Equality .0561
Family security .1931
Freedom .0453
Happiness .1063
Inner harmony .0781
Mature love .0562
National security .0667
Pleasure .0802
Salvation .0707
Social recognition .0348
Self-respect .0882
Ture friendship .0861
Wisdom .0906
Mean a .08478

*"N" refers to the number of individual MR's that determine
each entry; the number of subjects is 22.

-53-

Table 14. Mean MR (measure of relatedness) for instrumental
values, without defining phrases.

 

 

 

Values (N* 8 231)
Ambitious .1782
Broadminded .1156
Capable .1891
Cheerful .3160
Clean .1433
Courageous .2498
Forgiving .1767
Helpful .1163
Honest .2059
Imaginative .1238
Independent .1267
Intellectual .1457
Logical .0700
Loving .1096
Obedient .0595
Polite .1813
Responsible .1492
Self-controlled .0512
Mean = .15043

*"N" refers to the number of individual MR's that determine
each entry; the number of subjects is 22.

-59-

Table 15. Mean MR (measure of relatedness) for control words.

 

 

 

Control

Control words* Group
me = 2111

Fat .1258
Refrigerator .2513
Death .0735
Brother .0734
Thunderstorm .2355
Utensil .1024
Month .1429
Bride .0729
Smooth .0859
Tongue .1904
Apple .2359
Flag .1415
Orgy .1557
Brown .1484
Boy .1614
Mean .14646

*Words are listed here in the order in which they were
presented to the subjects.

** "N" refers to the number of individual MR's that
determine each entry; the number of subjects is 22.

-60..

Table 16. Summary of‘p tests comparing values to control words, for
associative response hierarchies and semantic differential.

 

 

Terminal Instrumental
Values Values
Associative Response Hierarchies:
Mean MR for values vs. control words: +3.607*** -0.176

Semantic Differential: comparison of
the amount of variance for three factors:*

Evaluation -2.135** -5.217****
Potency -0.048 -4.l31****
Activity +2.114 -3.397***

 

#5 tests were performed on the standard deviations.

** p .05
*** p .01
**** p .001

Positive‘p ratios represent greater similarity of meaning for the
control words, compared to the values. All tests are two-tailed;
degrees of freedom is 15.

:§£:.

terminal and instrumental values-~13 basically an evaluative process.
The high correlations (.68 and .62) with the semantic differential lend
a good deal of construct validity to the value scales. These correla-
tions seem especially high when they are considered in view of the
reliabilities of the terminal and instrumental value scales, which are
.71 and .66 respectively.9

Potency and activity, on the other hand, seem to be only minor
determinants of the ranking process for most subjects. The median
correlations which we did obtain for potency and activity (see Table
9, page 49) can be accounted for by chance. However, these correlations
do suggest two other possibilities which we can only mention here.
First of all, (as suggested earlier by Osgood g£_§1,, 1961) there
may be a preference for strong and active ways of life in our culture.
Thus with a different culture--or perhaps even with a different samplevi
of values--these correlations might vanish or even reverse themselves.
A second possbiility is that when subjects rank the values they respond
to the word "important" in the instructions. It may be that "important"
elicits responses which are not solely determined by evaluation.
Question 3

Question three was concerned with whether the associative meaning
of a value was different for people who ranked that value differently.
To test this possibility, we looked at the MR for subjects who ranked
a particular value either high (the upper quartile of ranks which a value
received) or low (the lower quartile of the ranks that a value received).

The between-quartile mean MR, instead of being lower than both the upper

 

9Rokeach, personal communication. The reliabilities are median Spearman
rho correlations with a seven-week retest.

-62-

and the lower quartile's mean MR, was lower than the upper quartile's
but higher than the lower quartile's mean MR (see Table 4 and 5).
What this means is that a subject in the lower quartile tends to give
responses which are more similar to subjects' responses in the upper
quartile than to responses from other subjects in the lower quartile.
Thus the associative meaning for the lower quartile may be said to
be similar to the associative meaning for the upper quartile, except
that for the lower quartile the meaning is more heterogeneous.

As we mentioned earlier, we have not yet explained the nature of
this greater heterogeneity of associative meaning for the lower quartile.
One explanation is that the associative meaning, strictly speaking,
did not become more heterogeneous for the lower quartile. Rather, the
lower mean MR for the lower quartile was a result of the evaluative
component of the connotative meaning of the values becoming more varied
between subjects.10 This explanation relies on the nature of the values.
It is assumed that all values ranked high by a subject are indeed positive
values to him. But values ranked low might be positive, neutral, or

even negative to a subject. Hence, to the extent that subjects let

their feelings toward the value determine their responses (see Appen-
dix D), the lower quartile could appear more heterogeneous (when in fact
there might be no more disagreement between subjects on the denotative
meaning of the values). If this explanation is correct, it should be
reflected by greater variance on the evaluative factor of the semantic
differential for the lower quartile as compared to the upper quartile.

In order to check this possibility,‘p tests were performed on the standard

10Of course, if subjects followed instructions exactly, the connotation
of the values should not affect their responses.

-63-
deviations of the semantic differential ratings for evaluation, potency
and activity. (The purpose of the tests on potency and activity was
to provide a comparison.) The difference in the variance (or standard
deviation) on the semantic differential ratings for evaluation, potency
and activity. (The purpose of the tests on potency and activity was
to provide a comparison.) The difference in the variance (or standard
deviation) on the semantic differential ratings was significant on
evaluation for both terminal and instrumental values (p = 2.68, p = .01,
and‘p = 2.21, p = .025 respectively). In each case there was greater
variance for the lower quartile. The.p tests for potency and activity
gave.p ratios from -.50 to +1.00; none approached significance. These
results support the hypothesis that the greater heterogeneity in
associative meaning for the lower quartile could simply be a result of
greater differences in evaluation. Naturally we cannot accept such
evidence as conclusive, since there are, no doubt, other possible
explanations.

Question 4.
Question four was concerned with whether we can say that in
general the values have the same meaning (associative and connotative)
to everybody. We have pointed out that meaning must be measured indirect-
ly, that similarity of meaning can be only relatively determined, and
that baseline or control data is necessary to make an answer interpretable.
There are two problems, however, with the comparison between the
values and the control words which should be mentioned here. First of
all, there can be some doubt that a variable which might be termed

"level of abstraction" has been adequately handled. Rokeach (1967)

-64-
points out that unlike attitudes, values are not specific to any
particular object or situation. In a sense, then, values have a type
of abstraction which the control words do not attain. The question is,
then, whether the variables on which the values and the control words were
matched (frequency of usage and meaningfulness) can be applied indis-
criminately to words which have different levels of abstraction.

The second problem with the comparison of the values to the control
words concerns the basic rationale behind such a comparison. Miller
(1951) has stressed the importance of context in communication. Words
receive a great deal of their meaning from the context in which they
appear. It may be that the control words themselves, taken in isolation,
do not have enough meaning similarity to justify most of the purposes
to which the value scales are put. However the value scales do provide
a type of context of their own, and more will be said of this later.

Outside of these two possible objections, theresults (see Table 16)
for the instrumental values need little discussion. 0n the semantic
differential the instrumental values show significantly greater similarity
of connotative meaning than the control words for all three factors.

For the associative response hierarchies there is no significant
difference between the instrumental values and the control words,and
the direction of the difference shows the instrumental values having
greater similarity of associative meaning.

For the terminal values, however, there is a complication. For
the associative response hierarchies the terminal values show significantly
less similarity of associative meaning than the control words. But

on the semantic differential there is no significant difference between

-65-
tenminal values and control words in connotative meaning similarity
for potency and activity; on evaluation the terminal values show signi-
ficantly greater similarity.

With the terminal values, in other words, we have found that the
referents vary from person to person (more so than the referents of
the control words), while the connotation is relatively stable. An
example of this for a value like "a comfortable life" would be that
to one person the value stands for security, an interesting job, and a
good car, while to another person it represents money and leisurely
travel. These are the referents of the value and we would expect a
low MR for these two people. However, both of them gay feel that "a
comfortable life" is a "strong" but "not-too-active" style of life.
Another example would be "an exciting life." Two people could agree
that it means a strong and active style of life, yet to one it may
mean parties and sports and to the other it may mean travel or danger.

The only negative implications which these results have for the
use of Rokeach's value scales concern the terminal values. The relatively
low degree of agreement in the associative meaning of these values
suggests that caution should be used in interpreting any differences
inthe value systems (as measured by the terminal value scale) of
different individuals.

However, the low degree of associative similarity for the terminal
value may be offset by two other factors. First of all, the explanatory

phrases which are added to the terminal values may be assumed to increase

-66-
inter-subject agreement on the associative meaning of the values.
Secondly-~and perhaps more importantly--the value scale in itself
provides a context for each of the values. That is, after a brief
examination of the values the subject can adopt a specific set toward
the type of concepts which he is rating.

This type of context may not only aid subjects in responding
to the different values, but may also aid the experimenter in inter-
preting the subjects' responses. Thus, in some sense, one can tell
what a person means by a value according to how he places the other
17 values. This is in accordance with Rokeach's (1967) conception
of a value system, in which the significance of any one value is
determined by its relationship to other values.

Nevertheless, when the use of the terminal value scale is
focused on one or two values, it would seem that as much care as
possible should be taken to determine the associative meaning of these
values to different subjects.

With respect to studying individual values, we have barely
touched on individual differences among the values. Tables 10 and 11
(pages 52 and 53) show that there are significant differences among
both terminal and instrumental values when measured (with their explana-
tory phrases) on their degree of meaning similarity. Since our concern
has beenvith general relationships which apply to all of the values, we
have ignored these results until now. This result suggests that a more
intensive investigation of particular values might be promising. With
the first value, for instance, we see that subjects in both the upper and

the lower quartile tend to agree that "a comfortable life" means money,

-67-

home, wealth and security. (See Appendix D.) The upper quartile,
however, feels that it also means contentment and leisure, while the
lower quartile tends to ignore this meaning. Of course with so many
possible responses this could easily be a chance difference and we

do not advise making much of this result at this time. But it suggests
that one could focus on this possibility to see if it might be the
major difference between the two groups. Then a further question would
have to be answered: do the subjects in the upper quartile feel that
money and wealth will result in contentment and leisure, or are they
saying instead that contentment and leisure are two of the things

one must have in order to have a confortable life? How would the lower
quartile have responded if the description of the value included the
idea of leisure and contentmert?

Questions similar to these can be formulated for most of the values.
To collect the proper data for answering them, more extensive measures
of meaning would be necessary--such as personal interviewing or content
analysis of lengthy descriptive definitions.

In conclusion, a word is in order about the generality of the
results of this study. The groups of subjects were relatively
homogeneous with respect to age, level of education, and academic
interests. The extent to which the results that were found would hold
for a more heterogenous cross-section of the population must still be

investigated.

REFERENCES

Barton, A. Measuring the values of individuals. Religious Education,
1962, £1, Research supplement, 62-97.

Bousfield, W. A. The problem of meaning in verbal learning. In
C. N. Cofer (Ed.), Verbal learning and verbal behavior. New
York: MeGraw-Hill, 1961.

Bousfield, W. A., Cohen, B. H., and Silva, J. G. The extension of
Marbe's law to the recall of stimulus-words. American Journal
‘2: Ps cholo , 1956, 92, 429-433.

Burt, C. The psychology of value. British Journal'pg Statistical
Psychology, 1963,‘1§, 59-104.

Campbell, D. T. Social attitudes and other acquired behavioral dis-
positions. In S. Koch (Ed.), Psychology: §;study p§.p_science,
Vol.'§. New York: MeGraw-Hill, 1963.

Carnap, R. The unity‘pf science. (Translated by M. Black.) London:
Regan Paul, Trench, Truber and Company, 1934.

Cattell, R. B. Personality and motivation structure and measurement.

Yonkers-on Hudson, N. Y.: World Books, 1957.

Creelman, M. B. The experimental investigation.p£_meaning. New York:
Springer Publishing Company, 1966.

Deese, J. The structure p§.association pp language and thought.
Baltimore: The Johns Hopkins Press, 1965.

Dollard, J., and Miller, N. Personality and psychotherapy. New
York: McGraw-Hill, 1950.

Duffy, E. A critical review of investigations employing the Allport-
Vernon Study of Values and other tests of evaluative attitude.
Psychological Bulletin, 1940,‘31, 597-612.

Dukes, W. F. Psychological studies of values. Psychological Bulletin,
1955, gg, 24-50.

Gage, N. L. Untitled review in 0. K. Buros (Ed.), The fifth mental
measurements yearbook. Highland Park, New Jersey: Gryphon Press,
1959.

Garskof, B. E., and Houston, J. P. Measurement of verbal relatedness:
An idiographic approach. Psychological Review, 1963, 1Q, 277-288.

Helson, H. Perception and personality-~A critique of recent experimental

literature. Randolph Field, Texas: Project report No. 1, Project
21-0202-0007. Air University, USAF, Sch. Aviat. Med. 1953.

-68-

-59-

Hill, W. F. Learning theory and the acquisition of values. Psychological

Review, 1960,.QZ, 317-331.

Jenkins, J. J., and Cofer, C. N. An exploratory study of discrete
free association to compound verbal stimuli. Psychological
Reports, 1957, 2, 599-602.

Kelly, G. A. The psychology of personal constructs. Vol. 1, (A
theory prpersonality. New York: Norton, 1955.

Lecky, P. Self-consistency: A_theory p§.personality. New York:
Island Press, 1945.

Loehlin, J. C. Word meanings and self descriptions. Journal pf
Abnormal and Social Psychology, 1961,‘p2, 28-34.

Loehlin, J. C. Word meanings and self-descriptions: A replication and
extension. Journal p£.Personality and Social Ps cholo , 1967,
5, 107-110.

MacCurdy, J. T. Psychopathology and social psychology. Part III:
Hierarchies of interest. British Journal pf.Psychology, 1950,
_4_1, 1-13.

Marshall, G. R., and Cofer, C. N. Associative indices as measures of
word-relatedness--A summary and comparison of ten methods.
Journal pf.Verbal Learning and Verbal Behavior, 1963, 1) 408-421.

Martin, W. E. Learning and identification: III. The development of
values in children. Journal‘pg Genetic Ps cholo , 1954, s3,
211-217.

Miller, G. A. Language and communication. New York: McGraw-Hill, 1951.

Morris, C. Paths pf_life. (Originally published, Harper, 1942.) New
York: George Braziller, 1956.

Morris, C., and Jones, L. V. Value scales and dimensions. Journal
p£_Abnorma1 and Social Ps cholo , 1955,‘§l, 523-535.

Noble, C. E. An analysis of meaning. Psychological Review, 1952,
_5_9_, 421-430.

Noble, C. E. Meaningfulness and familiarity. In C. N. Cofer and
B. S. MusgraVe Eds.), Verbal behavior and learnipg. New York:
MeGraw-Hill, 1963.

Noble, C. E., Stockwell, F. E., and Pryer, M; W. Meaningfulness (p')
and association value (p) in paired-associate syllable learning.

Psychological Re orts, 3, 441-452.

-70-

Osgood, C. E. Comments on Professor Bousfield's paper. In C. N.
Cofer (Ed.), Verbal learning and verbal behavior. New York:
McGraw-Hill, 1961.

Osgood, C. E., Suci, G. J., and Tannenbaum, P. H. The measurement'pg
meaning. Urbana: University of Illinois Press, 1957.

Osgood, C. E., Ware, E. E., and Morris, C. Analysis of the connota-
tive meanings of a variety of human values as expressed by
American college students. Journal of Abnormal ppg’Social
Psychology, 1961, pg, 62-73.

Piaget, J. The moral development péDthe child. Glencoe, Illinois:
Free Press, 1932.

Piaget, J. Will and action. Bulletin pf.pp§.Menninger Clinic, 1962,
gg, 138-145.

Rokeach, M. .A theory of organization and change within value and
attitude systems. Journal prSocial Issues. (In press.)

Rosenberg, M. J. Cognitive structure and attitudinal affect.
Journal p£.Abnorma1 and Social Psychology, 1956, 2;, 367-372.

Smith, M. B. Personal values as determinants of a political attitude.
Journal pf_Ps cholo , 1949, 28, 477-496.

Spranger, E. Types of men. (Translation of fifth German edition by
P. J. W. Pigors.) Halle: Niemeyer, 1928.

Thorndike, E. L., and Large, I. The teacher's word-book p£,30,000
words. New York: Bureau of Publications, Teachers College,
Columbia University, 1944.

Tolman, E. C. A psychological model. In parsons, T. and Shils, E.
(Eds.) Toward p_general theory p§_action. Cambridge, Mass.:
Harvard University Press, 1951, 279-361.

White, R. K. Value-analysis. Glen Gardner, New Jersey: Libertarian
Press, 1951.

Wickert, F. A test forpersonal goal-values. Journal p£.Social
Psychology, 1940,.ll, 259-274. (a)

Wickert, F. The interrelationships of some general and specific
preferences. Journal p§_Social Ps cholo , 1940, 1;, 275-302. (b)

Winthrop, H. The Pythagorean complex in the behavioral sciences.
Journal pf_Genetic Ps cholo , 1960,.2Q, 301-319.

Winthrop, H. Some neglected considerations concerning the problem
of value in psychology. Journal p£_General Ps cholo , 1961,
9.9;, 37‘590

Appendix A: Terminal and instrumental values used on
Rokeach's value scales.

I. Terminal Values

A comfortable life (a prosperous life)

An exciting life (a stimulating, active life)

A sense of accomplishment (lasting contribution)
A world at peace (free of wardand conflict)

A world of beauty (beauty of nature and the arts)
Equality (brotherhood, equal opportunity for all)
Family security (taking care of loved ones)
Freedom (independence, free choice)

Happiness (contentedness) ,

Inner harmony (freedom from inner conflict)
Mature love (sexual and spiritual intimacy)
National security (protection from attack)
Pleasure (an enjoyable, leisurely life)
Salvation (saved, eternal life)

Social recognition (respect, admiration)
Self-respect (self-esteem)

True friendship (close companionship)

Wisdom (a mature understanding of life)

II. Instrumental Values

Ambitious (hard-working, aspiring)
Broadminded (open-minded)

Capable (competent, effective)

Cheerful (lighthearted, joyful)

Clean (neat, tidy)

Courageous (standing up for your beliefs)
Forgiving (willing to pardon others)

Helpful (working for the welfare of others)
Honest (sincere, truthful)

Imaginative (daring, creative)

Independent (self-reliant, self-sufficient)
Intellectual (intelligent, reflective)
Logical (consistent, rational)

Loving (affectionate, tender)

Obedient (dutiful, respectful)

Polite (courteous, well-mannered)
Responsible (dependable, reliable)
Self-controlled (restrained, self-disciplined)

-71-

Appendix B: Median ranks assigned to values by subjects
for this experiment (N = 88).

 

 

 

. Range
Terminal Values Medipn High Lop_____
A comfortable life 12.83 18
An exciting life 8.50 18
A sense of accomplishment 5.95 17
A world at peace 9.75 18
A world of beauty 12.83 18
Equality 10.67 18
, Family security 11.00 18
Freedom 5.08 15
Happiness 7.79 17
Inner harmony 9.32 18
Mature love 5.17 18
National security 14.01 18
Pleasure 13.07 18
Salvation 17.28 18
Social-recognition 12.75 18
Self-respect 6.50 14
True friendship 8.30 18
Wisdom 5.23 18

 

-72-

-73-

Appendix B, continued.

 

Instrumental Values

 

Ambitious
Broadminded
Capable
Cheerful
Clean
Courageous
Forgiving
Helpful
Honest
Imaginative
Independent
Intellectual
Logical
Loving
Obedient
Polite
Responsible

Self-controlled

Range
Median High Low
10.68 1 18
4.25 1 17
7.41 1 17
12.23 1 18
14.21 3 18
8.50 1 17
10.64 1 18
10.30 1 18
3.67 1 16
9.00 1 18
6.50 1 18
6.93 1 18
9.95 l 18
7.93 1 18
16.90 1 18
14.06 5 18
5.06 1 16

9.93 1 18

 

Appendix C: Semantic differential instructions.

The purpose of this study is to measure the meanings of certain
values to various people by having them judge them against a series of
descriptive scales. In taking this test, please make your judgments on
the basis of what these values mean to‘ypp. On each page of this book-
let you will find a different concept to be judged and beneath it a set
of scales. You are to rate the concept on each of these scales in order.

Here is how you are to use these scales:
If you feel that the concept at the top of the page is very closely

related to one end of the scale, you should place your check-mark as
follows:

 

fair X : : : : : : unfair
0R
fair : : : : : : X unfair

 

 

 

 

 

If you feel that the concept is guite closely related to one or
the other end of the scale (but not extremely), you should place your
check-mark as follows:

: : weak

strong : X : :

 

 

OR

strong : : . : : X : weak

 

 

 

 

 

If the concept seems only slightly related to one side as opposed
to the other side (but is not really neutral), then you should check
as follows:

active : : X : : ' ° passive

OR

active : ° : X : : passive

 

 

 

 

 

The direction toward which you check, of course, depends upon which
of the two ends of the scale seem most characteristic of the thing you're
judging.

-74-

-75-

Appendix C, continued.

If you consider the concept to be neutral on the scale, both
sides of the scale egually associated with the concept, or if the scale
is completely irrelevant, unrelated to the concept, then you should
place your check-mark in the middle space:

: : : X : : :

 

 

 

 

 

 

IMPORTANT: (1) Place your check-marks in the middle pf spaces,
not on the boundaries:
THIS NOT THIS
: : X : :

:X

 

 

 

(2) Be sure you check every scale for every concept--
dp_not omit any.

(3) Never put more than one check-mark on a single scale.

Sometimes you may feel as though you've had the same item before
on the test. This will not be the case, so dp.ppp,look back ppg forth
through the items. Do not try to remember how you checked similar items
earlier in the test. Make each itemԤ_separate ppd independent judgpent.
Work at fairly high speed through this test. Do not worry or puzzle over
individual items. It is your first impressions, the immediate "feelings"
about the items, that we want. On the other hand, please do not be

careless, because we want your true impressions.

Appendix D: Ten most frequent associations given for each value.

Terminal Values:

 

 

 

 

 

 

 

 

A comfortable life An exciting life

Frequency* Frequency*
Associaplpn I, U L Association T U L
money 15 3 4 travel 17 5 4
home 11 2 3 interesting 7 l 2
wealth 10 4 1 fun 6 1 1
content 8 4 O challenge 6 3 0
security 8 1 3 adventure 5 2 1
leisure 7 3 0 people 5 1 1
car 5 l 2 happiness 5 O 1
free 5 l 1 moving 4 3 0
no worry 5 2 1 sports 4 1 0
clothing 5 l 1 social 4 l l
A sense of accopplishment A world at peace

Frequency* Frequency*
Association T U L Association T U L
recognition 11 2 4 love 13 4 4
pride 9 2 4 harmony 8 4 2
satisfaction 9 3 3 freedom. 7 2 O
respect 8 3 1 friendship 6 0 2
esteem. 6 2 1 brotherhood 6 2 1
fulfillment 5 2 0 happy 6 1 2
complete 5 l 2 fulfillment 5 2 0
success 5 1 1 equality 5 2 1
helpful 4 1 1 security 4 2 l
musician 4 2 O satisfaction 4 1 O
achievement 4 1 2 sharing 4 l O

' worth 4 O l quiet 4 1 3

work 4 O 2

 

 

*Number of times that the association was given by the total group
(T, N - 44), the upper quartile (U, N =11), and the lower quartile
(L, N - 11).

-75-

-77-

Appendix D, continued.

 

 

 

 

 

 

A world of beauty ‘ Eguality
Frequency* Frequency*
Association T U L Association I U L
painting 10 4 1 freedom. 15 4 5
tree 8 3 2 love 10 2 2
flower 6 1 2 friendship 8 1 3
alive 6 1 1 same 7 1 3
outdoor 5 3 0 integration 6 1 3
music 4 2 1 understanding 6 O 2
appreciation 4 1 1 peace 5 l 0
art 4 3 1 fair 5 l 1
freedom 4 l O Negro 5 0 2
serene 4 0 0 justice 4 2 1
right 4 1 0
Civil Rights 4 l 1
Family security Freedom
Frequency* Frequency*
Association T U L Association T U L
love 19 6 6 do-as-please 7 3 3
happiness 10 3 2 individuality 7 5 0
money 7 2 2 democracy 6 4 0
health 6 2 l equality 5 l 2
home 6 0 2 ’self-determination4 1 1
peace 5 2 1 responsibility 4 O 0
children 5 0 3 speech 4 2 0
wife 4 1 1 live 4 O 0
concern 4 1 3 different 3 2 O
shelter 4 O 2 happiness 3 O 1
decision 3 0 1

 

 

*Number of times that the association was given by the total group
(T, N - 44), the upper quartile (U, N - 11), and the lower quartile
(L, N - 11).

 

Appendix D, continued.

Happiness

Association

love
security
satisfaction
accomplishment
free

peace
friendship
fun

respect
money
recognition
wife

Frequency*

T

19
15
10

b-L‘J-‘J-‘J-‘U'Iflmm

U

C>c>hih>h>thahdh>$~$~a\

L

F‘P‘F‘P‘P‘F‘PIBJUDF‘UJUJ

 

Mature love
Association

understanding
giving
marriage
fulfillment
sharing

wife

empathy

mate

respect
considerate
happiness
security
responsibility
trust
kindness

care

helping
wisdom

Frequency*

T

11
11

wwwwwwwuwhkakbmo‘

U

HHOHOOOHNOHHNNDOWN

L

NOOOHNHHHHOHOOCwI-‘N

 

Inner harmony

 

Association

peace
happiness
content

secure

quiet
satisfaction
accomplishment
self-respect
confidence

no worries

Frequency*
T U L
15 4 3
7 4 l
6 O 2
5 2 1
5 1 O
5 3 1
5 2 1
4 2 l
4 l 0
4 0 3

 

National security

Association

defense
freedom
safety
government
peace
secure
army
military
force
unity
disarmament

 

Frequency*
T U L
10 2 1
9 3 l
8 l 5
5 l 1
4 3 O
4 1 1
3 2 1
3 l O
3 O l
3 l 1
3 0 O

 

*Number of times that the association was given by the total grou

the upper quartile (U, N - 11), and the lower quartile (L, N - 11 .

(I, N a 44):

, (+513

   

 

-79-

Appendix D, continued.

 

 

 

 

 

 

Pleasure Splyation

Frequency* Frequency*
Association T U L Association T U L
happy 13 7 2 heaven l4 5 3
fun 11 4 2 God 13 5 1
easy 6 2 1 religious 9 2 3
money 6 2 2 faith 5 2 0
love 6 2 0 peace 5 2 1
sex 5 1 1 belief 5 2 1
travel 5 1 1 hell 5 2 O
relaxing 5 1 1 church 5 2 0
comfort 5 1 3 Christ 5 1 1
satisfaction 5 O 1 after 4 2 0
Social recognition Self-respect

Frequency* . Frequency*
Association T U L Association T U L
friends 10 2 2 accomplishment 13 2 4
accomplishment 10 l 2 pride 7 3 2
esteem. 6 2 1 satisfaction 6 3 l
wealth 6 2 0 honest 5 l 1
respect 5 2 O confidence 5 2 2
successful 4 l 1 understanding 4 1 l
pride 4 O l admirable 4 1 O
looked-up-to 4 O 3 peace 4 O 1
position 4 l 1 happiness 4 l 1
prominent 3 2 1 worth 3 2 0
helpful 3 l O contentment 3 l 0
knowledge 3 O O inner 3 1 1
acceptance' 3 1 0 sincere 3 O 1
known 3 l 0 knowledge 3 O 1
leader 3 l 1 true 3 O 1
fame 3 1 1 harmony 3 0 1
status 3 l 1 consistence 3 l l

achievement 3 l l

 

 

*Number of times that the association was given by the total group
(T, N - 44), the upper quartile (U, N - 11), and the lower quartile (L, N - ll).

-30-

Appendix D, continued.

 

 

 

 

 

 

 

 

True friendship Wisdom
Frequency* Prequency*
Association T U L Association T U L
love 14 4 2 knowledge 20 2 5
trust 10 2 2 understanding 10 5 2
empathy 8 1 4 intelligence 6 2 2
respect 8 l 3 truth 6 O 2
helpful 7 l 2 insight 6 1 2
depend 4 O 1 peace 6 l 1
sharing 4 l O openminded 5 2 1
giving 4 O 2 experience 5 1 2
fulfilling 4 l 0 education 5 1 2
concern 4 1 l awareness 4 O 1
Instrumental Values:
Ambitious pp, Broadminded
Frequency* Frequency*
AggggiggiOQ. T U L Ass a T U
goals 11 3 1 liberal l6 4 6
success 7 2 2 understanding 12 2 1
striving 7 2 2 fair 10 2 3
driving 6 1 1 listens 7 2 1
energetic 5 3 1 accepting 7 O 5
pushy 5 O 2 free 6 2 0
dedicated 5 l 2 unprejudiced 6 1 2
hard 5 O 2 unbiased 5 0 1
money 5 O 3 rational 4 2 0
free 4 1 0 open 4 l 0
logical 4 1 O perceptive 4 l 0
logical 4 l 0
responsible 4 0 O

 

*Number of times that the association was given by the total group (T, N a 44),
the upper quartile (U, N - 11), and the lower quartile (L, N - ll).

Capable

Association

able
efficient
reliable
successful
good

working
responsible
knowledgeable
skillful
worthy

Appendix D, continued.

Frequency*

T

12
12

UIU'IUIO‘O‘O‘NQ

U

ONNOHWWUOWUJ

L

wOHr-HHHOHUI

 

Clean
Association

order
health
spotless
well-groomed
wash
meticulous
shiny
fresh
organized
bright
fastidious
sanitary

Frequency*

T

H
wwab-P‘UIMGGQJ-5

U

HONOHNNNNNNW

L

HOOHNHNOHHU-k

 

*Number of times that the association.was given by thetotal group (T, N - 44),
the upper quartile (U, N - 11), and the lower quartile (L, N - 11).

-31-

Cheerful

Association

happy
smiling

friendly
carefree
83?

fun

kind
easy-going
loving
pleasant
optomist

Frequency*

T

on
O

H
UIU'IUIUIUIUIQCDQO

U

1

HHHHHUNUUONH

L

UNOOOO-I-‘NNOO‘

 

 

Courageous

Asso ation

brave
strong
bold
unafraid
honest
heroic

guts
independent
loyal

sure

Frequency*

T

15

ll

whbbmumoo

U

HHNNNDJHNUIU

L

F‘F‘C’F‘P‘F‘F‘U’U’U‘

 

-oo-

- - -
o c
- .

 

   

 

 

 

w-

o

-32-

Appendix D, continued.

 

 

 

 

 

 

 

 

 

 

Forgiving Helpful
Frequency* Frequency*
Association T U L Association T U L
understanding 20 5 0 kind 12 4 2
kind 12 3 3 concerned 10 2 l
loving 9 3 l unselfish 7 2 2
sympathy 5 2 2 giving 0 2
forget 5 3 1 loving 5 3 O
tolerant 5 1 1 friendly 5 3 2
open-minded 5 1 l useful 5 2 O
God 4 1 1 understanding 5 1 0
helpful 4 1 1 caring 4 2 0
compassionate 4 1 2 willing 4 1 0
humanitarian 4 0 1
generous 4 O O
self-sacrificing 4 O 2
Honest Imaginative
Frequency* Frequency*
Assogiation T U L Associat en T U L
trustworthy 14 2 2 artistic l3 2 4
frank 7 2 3 thinking 9 2 1
open 5 2 0 original 5 2 0
straightforward 5 l 0 free 4 2 1
fair 4 2 1 expressive 4 2 l
dependable 4 O l broadminded 4 1 2
courageous 4 O l clever 4 0 2
true 3 2 O intelligent 4 l 1
candid 3 2 1 independent 3 1 O
reliable 3 2 0 bold 3 1 O
responsible 3 l 0 exciting 3 l O
loyal 3 1 O innovative 3 1 0
real 3 O O unusual 3 l 0
independent 3 1 O idea 3 l O
honorable 3 0 0 different 3 1 0
free 3 O 1 sharp 3 O 1
good 3 1 l witty 3 1 l
liberal 3 l l
open-minded 3 O 2

 

*Number of times that the association was given by the total group
(T, N - 44), the upper quartile (U, N I 11), and the lower quartile
(L, N - ll).

Appendix D, continued.

 

 

 

 

 

 

 

 

 

 

Independent Iptellectpgl
Frequency* Frequency*
Association T U L Association T U
alone 12 3 3 smart l4 2 4
free 8 l 2 knowledgeable 12 2 3
confident 6 3 2 thoughtful 7 2 2
strong 6 2 1 well-read 6 1 0
capable 6 2 1 thinking 6 1 0
responsible 5 3 0 interested 5 3 0
secure 4 2 0 logical 5 1 2
stable 4 l O educated 5 1 3
competent 3 2 0 learn 3 2 0
individual 3 1 1 open-minded 3 2 0
stand (alone) 3 O 0 books 3 1 1
wise 3 1 1
philosophy 3 O 2
creative 3 O 3
Logical Loving
Frequency* Frequency*
Asgocigtion T U L Association T U L
reason 5 2 0 kind 14 5 3
thought 5 2 O warm. 8 3 2
orderly 5 O 1 giving 8 2 1
intelligent 5 0 0 happy 7 4 1
sensible 4 l 0 understanding 6 O 1
clear 4 l l emotional 5 l l
mathematical 3 1 O respect 5 2 2
organized 3 1 O gentle 4 2 1
open-minded 3 O O caring 4 1 O
(15 words tied with 2 occurences) thoughtful 4 l 1
sweet 4 O 1

 

 

*Number of times that the association was given by the total group
(T, N - 44), the upper quartile (U, N a 11), and the lower quartile
(L, N - ll).

Appendix D, continued.

 

 

 

 

 

 

 

 

 

Obedient Polite
Frequency* Frequency*
Assogiggion T U L Association T U L
dependent 7 O 3 respect 14 4 4
respecting 5 4 O kind 7 3 1
dependable 5 l 0 others 5 3 l
authority 5 l 1 nice 5 l 1
do-as-told 4 1 1 considered 5 1 l
responsible 4 2 l upbringing 4 O 2
trust 3 2 l thoughtful 3 l 0
listen 3 l O well-meaning 3 l O
accepting 3 0 0 social 3 l 1
orders 3 1 l gentlemanly 3. l 0
superficial 3 O 1
friendly 3 l O
formal 3 O 3
obedient 3 O 1
Responsible Self-controlled
Frequency* Frequency*
Association T U L Association T U L
trustworthy 16 6 2 logical 6 l 2
capable l3 4 0 even-tempered 5 O 2
honest 7 2 l quiet 5 O 2
respected 6 2 3 inhibited 5 0 2
mature 6 l l unemotional 4 2 l
thoughtful 4 1 l knows self 4 2 l
obedient 4 2 1 independent 4 1 1
loyal 4 O 3 trained 3 l 0
independent 3 O l rational 3 O 1
confidant 3 O 1 mature 3 O 1
hard-working 3 O 1 responsible 3 O 1
effective 3 l 1
consistent 3 l 1
dutiful 3 1 2
disciplined 3 0 2

 

*Number of times that the association was given by the total group
(T, N - 44), the upper quartile (U, N a 11), and the lower quartile
(L, N - ll).

"11171953

HICHIGQN STQTE UNIV. LIBRQRIES

31293102378571