. I“ \ Ml

.— .n I ‘ I V i, .

:? vy‘fixzv ”Ki?“ “-3

X'M‘Nmuflntm‘qx ‘ - "p.
\ u ‘ a?“ I.) u.‘ L [\‘U, u ‘ Al") iris. 3

\¢\ ' <
w».

in: --““
.w.

a
' n
a. u'

A, . .’
.7 :1, F“, ~.~
.“ ' ‘
I . "
fix

~. - "

ii
a.-
fi'.“

. v
. “
‘ 33‘: ‘V‘19z4
4 A {le ‘ ‘ ,‘u. 9 l . w ‘ ‘ :.\' A‘I\
A . ‘31 - ‘ . , \vfllk.
‘\“"‘q~r‘. \> . a.“
. 5*

33s

_ M \f.
l}.\.‘_“&“* ‘
v ‘;‘\-' 7
x ""5-

3‘.

g
M,
L

\

)n. '
L.\" e“-
‘ «TA-
..- q‘x

 

(5 EMMMv. _
3:1“? ,. , (v
MKh’T‘fifi 373,-;

U?" "731%; .v'
m ‘J _’$ (5.1 'I

im- r»..- ...,

    

:",I

u‘a

This is to certify that the
thesis entitled
THE DEVELOPMENT AND ARTICULATION OF

ATTRIBUTES IN PERSON PERCEPTION

presented by

Daniel M. Wegner

has been accepted towards fulfillment
of the requirements for

Ph.D. Psychology

degree in

 

 

4122/ 9

Major professor

 

 

Date August 19, 1974

 

0-7639

 

 

 

&23
to L

fevej
.I.e 1

“it:

I

,.

=;:“q“
.‘cd

ABSTRACT

THE DEVELOPMENT AND ARTICULATION
OF ATTRIBUTES IN PERSON PERCEPTION

BY

Daniel M. Wegner

This research focused on the development and articulation of the
attributes given by individuals in describing others. Of special inter-
est was attribute utility, the generality with which the individual uses
an attribute to characterize stimulus persons. Experiment 1 was designed
to test the hypothesis, derived from cognitive-developmental theory, that
the utility of attributes employed by perceivers increases with the
development of the perceiver. It was also expected that differentiation,
the total number of different attributes given by a perceiver, increases
with the age of the perceiver.

Free descriptions of eight acquaintances were obtained from 88
subjects; the sample included approximately equal numbers of male and
female fourth, sixth, ninth, and twelfth grade students. These descrip-
tions were content analyzed such that the number of different attributes
and the utility of every attribute used by a subject could be determined;
low utility attributes were those used to describe only one of the
acquaintances, while high utility attributes were those used to describe
more than one of the acquaintances.

Preliminary analyses of the Experiment 1 data revealed that while
differentiation was positively related to subjects' verbal intelligence,
the utility ratio (the proportion of high utility attributes to total
number of attributes) was not. A sex by grade unweighted means analysis

of covariance of differentiation scores, with verbal intelligence as the

 

.
“"1
dfis‘

b‘.

~..€

Daniel M. Wegner
covariate, indicated significant main effects for sex and for grade.
Females displayed greater differentiation than males; differentiation
increased with age of subject. A sex by grade unweighted means analysis
of variance of utility ratios revealed a significant main effect for
grade; utility ratio increased with age of subject. In sum, the results
indicated that the number of different attributes identified by the per-
ceiver increases as the perceiver develops, and that the prOportion of
attributes having high utility also increases developmentally.

Experiment 2 was undertaken to elucidate the function of the
observed developmental increments in attribute utility. An informal
model of cognitive structure, synthesizing elements of previous categori-
cal and dimensional models, was introduced to facilitate the conceptuali-
zation of the relationship between attribute utility and attribute
articulation. It was hypothesized that the articulation of an attribute,
the likelihood that fine discriminations among stimulus persons are made
in terms of the attribute, is a concommitant of the level of utility of
the attribute. Samples of five high utility and five low utility attri-
butes were obtained from each of ten male and ten female undergraduate
subjects; the free description and content analysis procedures used in
Experiment 1 were employed for this purpose. Using two independent mea-
sures of attribute articulation, for which subjects were required to rate
and sort stimulus persons on attribute dimensions, it was found that sub-
jects' judgments were more articulated when based on high utility than
low utility attributes. It was proposed, in light of these results, that
the development of person perception is best interpreted in terms of the
increasing utility of attributes; such gains in utility enhance the prob-

ability that fine discriminations among stimulus persons may be made.

 

 

r--

Daniel M Wegner
The implications of the present findings for cognitive-
developmental theory and for future studies of person perception in chil-

dren and adults were discussed.

THE DEVELOPMENT AND ARTICULATION

OF ATTRIBUTES IN PERSON PERCEPTION

BY

Daniel M. Wegner

A DISSERTATION

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

DOCTOR OF PHILOSOPHY
Department of Psychology

1974

 

 

(—1

ACKNOWLEDGMENTS

This research was carried out under the guidance of Dr. William
Crano. His many substantive contributions to this work are greatly
appreciated.

The thoughtful comments of Dr. James Phillips and Dr. Robin
Vallacher also helped to shape the dissertation. Both deserve thanks for
their assistance. In addition, Dr. Ellen Strommen and Dr. Lawrence
Messé, who served as members of the committee, made many helpful

suggestions.

ii

 

-..
. 7“

-
~¢A.~

”e

s.

I']
m
73.! (n 3.. 3» (n

“V

I‘ P‘
.\ .

U‘JE
...
L‘s
U“‘
I
fit‘

TABLE OF CONTENTS

LIST OF TABLES
LIST OF FIGURES . . .
INTRODUCTION . . . . . . . . . . . . . .

The Structure of Person Perception . .
Core Categories: The Utility of an Attribute
The Internal Structure of Attributes . . . . .
Attribute Utility and Articulation . . .

The Development of Person Perception .
Attribute Utility and Cognitive Development
The Development of Person Perception: Previous
Summary and Hypotheses

EXPERIMENT 1

Method . . . . . . . . . . . . . . . . . . . . .
Subjects . . . . . . . . . .
Free Description Procedure . . . . . . . . . .
Coding System . . . . . . .
Coding Procedure and Intercoder Agreement
Design and Measures . . . . . . . . . . . .
Results . . . . . .
Preliminary Analyses
Differentiation
Utility Ratio
Summary

EXPERIMENT 2

Method . . . . . . . . . . . . . . . . . . .
Subjects and Design . . . . . . . . .
Attribute Selection Procedure . . . . . . . .
Articulation Measurement Procedure . . .

Results . . . . . .

Relationship Between Articulation Measures
The Effects of Attribute Utility . . . . . .

DISCUSSION .

Differentiation and Development . . . . . . . .
Attribute Utility and Development .

The Utility and Articulation of Attributes
Conclusions

iii

Research

vi

H

22

22
22
22
24
26
26
28
28
29
31
32

33

33
33
34
35
37
37
38

4O

4O
42
45
48

APPENDICES:

A. Free ReSponse Description Coding System .
B. Preliminary Analysis for Experiment 1

C. Analyses of Differentiation Scores

D. Analysis of Utility Ratios

REFERENCES .

iv

50

55

56

57

58

‘\

LIST OF TABLES

Table
1. Sample Sizes and Mean Ages (Years, Months) by Grade and Sex .

2. Original and Adjusted Cell Differentiation Means by Grade
and Sex . . . . . . . . . . . . . . . . . . . . . . . . .

3. Utility Ratio Cell Means by Grade and Sex . . . . . . .

4. Means of Within-Subject Means on Number of Groups and
Information Measures of Attribute Articulation for High
Utility and Low Utility Attributes . . . . . . .

Bl. Means and Standard Deviations of Variables Assessed in
Experiment 1 . . . . . . . . . . . . . . . . .

B2. Correlations Among Variables Assessed in Experiment 1 . .

Cl. Summary of Sex by Grade Unweighted Means Analysis of Variance
of Differentiation Scores . . . . . . . . . . . . . . . .

CZ. Summary of Sex by Grade Unweighted Means Analysis of Covari-
ance with Differentiation Scores as the Criterion Measure
and Verbal Intelligence Scores as the Covariate . . . .

D1. Summary of Sex by Grade Unweighted Means Analysis of Variance
of Utility Ratios . . . . . . . . . . . . . . . . . . . .

Page

23

29

31

38

55

55

56

S6

57

LIST OF FIGURES

Figure Page
1. Locus of Attributes Representing One Object . . . . . . . . . 9
2. Locus of Attributes Representing Two Objects (E and 01) . . . 10

3. Locus of Attributes Representing Four Objects
(E, 01, 02, O3) . . . . . . . . . . . . . . . . . . . . . . 10

4. Alternate Ordering of 01 and 02 with Respect to E . . . . . . ll

5. Locus of Attributes Which Preserve the Order (E, 01, 02) . . 12

vi

 

 

al

8t

INTRODUCTION

The Structure of Person Perception

The process whereby people perceive and come to understand other
people has been the focus of considerable theoretical and empirical
attention in recent years. Of the variety of empirical generalizations
emanating from this tradition, probably the most central is the notion
that the recognition of the particular attributes of a person (e.g., his
attractiveness, hair color, aggressiveness, and so on) is more a function
of the perceiver than of the person being perceived (Dornbusch et al.,
1965). The proverb "Beauty is in the eye of the beholder" is an apt sum-
mary of the often replicated finding that there is greater commonality
among reports on many people given by a single perceiver than there is
among reports on a single person given by many perceivers. An immediate
implication of this result is that ". . . a person has a core of gener-
ally consistent categories used in describing all people. . . . [Hastorf
et al., 1958, p. 61]." An understanding of the process through which the
individual deve10ps this core of categories is the primary concern of the

present investigation.

Core Categories: The Utility of an Attribute

The common assumption in most studies of person perception proc-
esses is that the perceiver's judgments of a person are a function both
of the person's characteristics and the perceiver's past experience with

people. Bruner and Tagiuri (1954) have suggested that this interaction

2

be considered in terms of the perceiver's "implicit personality theory,"
a set of assumptions the perceiver has created on the basis of past expe-
rience, which he employs in the construal of new experiences with people.
Perceptual and cognitive psychologists such as Gibson (1963) and Neisser
(1967), who emphasize the "constructive" aspects of perception and cogni-
tion, have pointed out that the process of constructing a cognition--in
this case, a cognitive representation of a person--leaves behind a resi-
due, a record of the construction process which guides the construction
of later cognitions. It is obvious that in forming a cognitive structure
for processing information about people, we do not keep representations
of entire persons in our heads. The observation that the mention of a
person as a "redhead" conjures up an entire structure--a stereotype of
what redheads are like--suggests that we process information about people
in smaller units. For many theorists, these units of interpersonal cog-

nition are the attributes of the stimulus person (Zajonc, 1968); an

 

attribute is operationally defined as a descriptive characteristic
attributed to a person (e.g., smart, nice to be with, wears glasses,
female, doesn't like small children). In other words, anything a per-
ceiver says about a person is an attribute of the person for that
perceiver.

In describing a particular individual, a perceiver probably
refers to some attributes which are very specific to that stimulus per-
son; it is unlikely, for example, that a perceiver would say "he wears
blue shoes" about very many of his acquaintances. On the other hand,
attributes may be very general; a perceiver might well point out that
many of the people he knows are ”considerate.” For the purposes of this
discussion, the generality with which a perceiver uses an attribute to

characterize stimulus persons will be termed the utility of the

3
attribute. One attribute has higher utility than another if the per-
ceiver employs it in the construal of a wider variety of stimulus per-
sons. By this definition, a "high utility attribute" corresponds closely
to the notion of a "core category" proposed by Hastorf et a1. (1958). .A
study by Faguy-Coté (1965) has provided evidence in support of the util-
ity definition of core categories. Using a variety of instruments, Faguy-
Coté isolated the high utility attributes used by each of her subjects.
These instruments included both adjective checklists, in which utility
was defined as the frequency of usage of an adjective across a variety of
stimulus persons, and a free response description task, in which the sub-
ject's high utility attributes were defined as those he gave as the
"essential characteristics" of the largest variety of stimulus persons.
In addition to finding a close correspondence between the two sets of
high utility attributes elicited from each subject in these tasks, she
also found that subjects' ratings of the "importance" of attributes were
significantly correlated with utility. These results suggest not only
that attribute utility can be unambiguously and reliably assessed, but
also that subjects agree with such an operational definition of a "core

category."

The Internal Structure of Attributes

 

An attribute is necessarily a judgment of a stimulus person; it
is not a pure assessment of the person, but rather the perceiver's
response to the person as a stimulus. As has been noted, many theorists
Choose to think about attributes as the product of an interaction between
the immediate input--a person--and the cognitive structure of the per-
ceiver. The assumption about the nature of the interaction that has been

implicit in the present analysis is that the judgmental process can best

 

be

th-

0?

4
be conceptualized as an act of categorization. Congruent with a number
of earlier formulations (Bruner, 1957; Gibson, 1963; Wallach, 1958), this
approach includes the supposition that the perceptual inputs to which the
perceiver is subject only become translated into potential responses when
they have been cognitively associated with some class of similar objects
or events. The basis of similarity is the recognition that all the
objects or events in a particular category have some attribute in common.
As Bieri et a1. (1966) have pointed out, this supposition implies that
cognitive structures--categories--are nominal and qualitative in nature,
and may be represented mathematically using the language of set theory.

A stimulus object is processed by placing it in a set or variety of sets
of similar objects, at which time it may be responded to on the basis of
the attributes defining these sets. If a person is standing on a rail-
road track, for example, and a speeding train is approaching, unless this
percept is translated into the concept of "Speeding train" through the
process of matching the input with the cognitive category associated with
Speeding trains, the person will have no basis for responding to the per-
cept, and will probably not live to worry about it.

The categorical model of judgment, however, seems difficult to
apply to situations in which perceivers are asked to respond to stimuli
according to their relative magnitude along some continuum. Individuals
are quite capable of perceiving and responding to differences in, say,
the brightness of a light, or the relative intelligence of a person.
Attneave (1962), Bieri et a1. (1966), and Restle (1961) have all recog-
nized that many judgments people make can be viewed intuitively in terms
of dimensional model of judgment. That is, the process of perceiving can
be seen as the process of assigning an input some value along an attri-

bute dimension; the perceiver bases re3ponses upon this value. In

S
contrast to the set-theoretic representations given to the categorical
model, the dimensional model is most often represented in terms of an
Euclidean n-Space in which attributes are the dimensions of the Space,
objects of cognition are points in the space, and the projections of each
object on the various dimensions represent the values or magnitudes of
the object's dimensional attributes (see, e.g., the models presented by
Scott, 1969; Rosenberg & Sedlak, 1972). However, since the data reduc-
tion techniques (i.e., factor analysis and multidimensional scaling)
typically used to represent dimensional cognitive structure hold the cen-
tral assumption that every object is construed on every attribute (cf.
Coombs, Dawes, & Tversky, 1970), the dimensional model denies the intui-
tively obvious idea that some attributes are apprOpriate to only a few
cognitive objects. It is difficult to conjure up a situation, for
example, in which one would think seriously about the happiness of a
crankshaft, the roundness of a square, or the amplitude of a window box,
yet this assumption is a necessary part of the present state of concep-
tualization of dimensional judgments.

Various researchers have attempted to reconcile the differences
between these two models of judgment, meeting in the process various
levels of success. Both Attneave (1962) and Restle (1961), for example,
discussed in detail the possible approaches to the representation of
dimensions in terms of sets, but concluded that such representations
required "stretching" mathematical set theory. A more promising avenue
of reconciliation between the two approaches is suggested by evidence
demonstrating that a category is not, in practice, a purely nominal and
qualitative distinction. Rosch (1973) has pointed out, for example, that

some colors to which English Speakers apply the word "red" are

"redder” than others. Some breeds of "dog" (such as retriever) are
more representative of the "meaning" of "dog" than are others (such

6

as Pekinese). . . . However, psychological and linguistic research
has tended to treat categories (whether perceptual or semantic) as
though they were internally unstructured-~that is, as though they1
were composed of undifferentiated, equivalent instances--and as
though category boundaries were always "well defined" . . . [p. 111].

On the basis of subjects' judgments of "goodness of fit” of objects in 2
categories, and subjects' reaction times to problems requiring the cate-
gorization of objects, Rosch concluded that categories of objects are
internally structured, and may be conceptualized as consisting of an
exemplar--an object which is the "best example" of a category--and other
objects which are less well represented by the category. In light of
these findings, it may be concluded that categories do not necessarily
represent homogeneous sets of objects, and are therefore not amenable to
mathematical set-theoretic explanations in any strict sense. These
results may also be taken, however, as support for a more general proposi-
tion: An attribute may be thought of as a dimensional category; the
objects in a category are organized along a dimension of "best fit," with
the exemplar standing at one extreme of the dimension.

This analysis points to a conclusion similar to that proposed by
Bieri et a1. (1966), that categorical judgments involve discriminations
among cognitive objects according to whether or not they are both con-
struable on the same attribute, and that dimensional judgments involve
discriminations among objects' values along an attribute. This second
type of judgment requires, of course, that both objects have already been
categorized on the attribute. However, the conclusion of Bieri and his
colleagues may be strengthened by noting that the process of judgment
involving low utility attributes is necessarily categorical, since only
one or a few objects are concerned--in contrast, high utility attribute

judgments may be categorical (i.e., "Does the attribute apply?") or

dimensional (i.e., "How close is the object to the exemplar?"). This

0r
Cat.
Obie

the

7
implied flexibility of high utility attribute judgments may be analyzed
in greater detail upon consideration of the relationship between the
utility of an attribute and the extent to which that attribute approxi-

mates a dimension.

Attribute Utility and Articulation

Given the assumption that attributes may be viewed as "dimen-
sional categories," distinctions among them may be made on the basis of
their level of articulation. Bieri (1966) defined this preperty of an
attribute as ”the number of categories or intervals which may be dis-
criminated along the dimension [p. 21]," and Scott (1969) arrived at a
similar formulation, that the articulation of an attribute is "the number
A of reliable distinctions among objects that a person makes on the attri-
bute [p. 263]." In a sense, then, the articulation of an attribute is
indicative of whether the attribute is better represented as a category
or a dimension. A completely unarticulated attribute is similar to a
category, since reliable discriminations are not made among members; each
object construed in terms of the attribute is consequently an exemplar of
the attribute, and discriminations among objects are categorical. A
highly articulated attribute, in turn, is one on which many reliable dis-
criminations among objects can be made, and is therefore amenable to
dimensional representation.

At the simplest level of analysis, it is obvious that the utility
and the articulation of an attribute are perfectly related, since an
attribute which applies to only one object cannot be articulated--only a
categorical discrimination may be made. One which applies to many
objects has the capacity for articulation. Scott (in press) has proposed,

on the basis of significant correlations between individual difference

8
measures of attribute utility and attribute articulation that ". .
refined discrimination along the attributes of appraisal tends to encour-
age their application to a large number of objects [p. 25]." Although
Scott has no empirical ground upon which to infer such a causal relation-
ship, his inference that the utility and articulation of an attribute are
somehow linked--beyond the simple level of analysis suggested here--
deserves careful consideration.

It is possible to construct an informal model of cognition in
which the relationship between utility and articulation may be examined.
Suppose, for example, that attributes can be represented as line segments
in a multidimensional space, and that objects of cognition are points in
the space. Assume further that the exemplar of an attribute is one end-
point of the line segment, and that objects which are construed on the
attribute are all the points in the space which have normal projections
on the line segment. It is possible to think of that line segment, then,
as being defined by the objects which have projections on it. In a sense,
each new object added to the attribute specifies more exactly where the
attribute must lie in the space, and therefore increases the probability
that any two objects on the attribute will be reliably discriminated.
Only when an attribute "holds still" in the space will the order of pro-
jections of points be stable enough to allow for articulation of judgments
on the attribute.

Consider a specific illustration to clarify this argument. Sup-
pose that an individual has a two-dimensional cognitive space, and that
there exists in the space a single attribute line segment with its cor-
responding exemplar. In this case, the attribute could be any line seg-
ment originating at the exemplar (illustrated in Figure 1). With the

addition of a single object (01) construed on the attribute, the locus of

 

Figure l. Locus of attributes representing one object.

line segments is halved. The requirement that an object construed on an
attribute must have a normal projection on the attribute eliminates the
set of line segments for which no normal projection exists. The attri-
bute defined by an exemplar (E) and one object (01), as shown in Figure
2, must lie to the right of line X (a line perpendicular to a line join-
ing E and 01). When, as shown in Figure 3, even more objects are con-
strued on the attribute, the locus of possible attributes decreases
further, since each new object restricts the locus in the same way it was
restricted with the addition of the first (01). According to this
example, it appears that the locus of potential attributes is a decreas-
ing function of the number of objects projected on the attribute (pro-
vided that new objects do not fall on lines between E and existing
objects--in which case their contribution to the definition of the attri-

bute is redundant).

lO

 

 

X

Figure 2. Locus of attributes representing two objects
(E and 01).

 

o 03
I,
/ X2
I
I
I
I
l
I
E o 01
\ fl
\
\
\
\
\
\
\
\
\
b
0
2 X3

Figure 3. Locus of attributes representing four objects
(E, 01’ 02, 03) o

11
The stability with which an attribute is located in the two-
dimensional space is the determinant of the attribute's articulation. If
the articulation of an attribute is defined as the reliability of ordinal
object judgments on the attribute, then a decreasing locus of possible
attribute line segments increases articulation. Consider the case of a
line segment which has an exemplar (E) and two other objects (01 and 02).

As shown in Figure 4, it is possible for the line segment to represent

0 01

Figure 4. Alternate ordering of 01 and 02 with respect to E.

the order of the objects as (E, 01, 02) or as (E, 02, 01). For the order
of objects to be consistently given as (E, 01, 02), only a restricted set
of attribute line segments may be used. As shown in Figure 5, the only
line segments which fulfill this requirement are those bounded by line A

(a perpendicular to the line connecting E and 01) and line B (a

 

per
clo
lin
md
to

of

O
V?!
I“?

12

Figure 5. Locus of attributes which preserve the order (E, 01, 02).

perpendicular to a line connecting 01 and 02). Any line segment counter-
clockwise past line A would not have a normal projection from 01; any
line segment clockwise past line B would have projections frouiboth 01
and 02, but would result in a reordering of the two objects with respect
to E (i.e., E, 02, 01). An articulated judgment, in which the ordering
of objects is held constant, can only be based upon a restricted locus of
attributes. Since increments in the utility of an attribute serve to
restrict its locus, a high utility attribute--according to the present
analysis--is more likely to be articulated than is a low utility
attribute.

The informal model and supportive arguments, although not proof
of a general relationship between utility and articulation, are redolent

of the direction such a proof might take. Yet, however the relationship

 

(7‘
as

ate

Early

C0231

13
between utility and articulation is eventually conceptualized, it is
essential to recognize that it implies a very Special function served by
the "core categories" employed by the individual. That is, a high util-
ity attribute, as contrasted with a low utility attribute, is more simi-
lar to a dimension of judgment than to a category, and as a dimension, is
more likely to provide the perceiver with a stable reference scale upon

which reliable discriminations among stimulus persons may be based.

The Development of Person Perception

 

Two relatively discrete linescflfevidence can be traced to the con-
clusion that the development of person perception is best conceptualized
in terms of the increasing utility of attributes. One, the cognitive-
developmental research dealing with the child's successive cognitive
representations of reality, provides evidence on person perception only
by analogy, since research on physical object cognition is not immedi-
ately applicable to questions on the nature of person perception. The
second, the research on the development of person perception, per se, is
also somewhat distant from the present concern because previous investi-
gations have treated attribute utility as a methodological problem and
not as a variable of theoretical interest. Despite these difficuties, it
is reasonable to suggest that cognitive-developmental theory provides a
general perspective for viewing attribute utility, and that some specific
findings may be isolated from the body of deve10pmental person perception

research that will serve to clarify this perspective.

Attribute Utility and Cognitive Development
Many theoretical treatments of cognitive development suggest that
early cognition is directly tied to perceptual events, and that later

cognition is more abstract, inferential, and generalized. It can be

 

u.
(r!
I.

proc
ind

art:

of 1
? ?w

h A;

than

Ceptx
as tl

from

not (
Enact
It is
bltes
Gilli
to th.
I“ Cor
Struct

CEiver

l4
argued that the dimension of attribute utility is the analogue of this
process in the realm of person perception. A low utility attribute is,
indeed, directly tied to a perceptual event, whereas a high utility
attribute represents a type of integrative theme wherein many objects of
cognition are grouped. Evidence for this developmental hypothesis can be
seen in the work of Jean Piaget, since as Wohlwill (1962) has pointed

". . . much of Piaget's work--particularly that on the conservation

out,
of length, weight, volume, number, and so forth--is interpretable in
terms of the increasing stability of concepts in the face of (irrelevant)
changes in the stimulus field [p. 87]." Continuing in this vein, Wohl-
will goes on to suggest that the major dimension of psychological growth
represented by Piaget's findings is that of perception--inference. In
essence, it is posited that early concepts, whether they are verbalized
or are preverbal, are based directly on perceptual experience. Such per-
ceptual concepts are by nature very unstable, appearing and disappearing
as the immediate environment changes. Later concepts are distinguished
from earlier ones on the basis of the formation of a COgnitive struc-
ture--the individual may make inferences about his present experiences
not only with regard to the character of these experiences, but also as a
function of the structures through which these inputs are represented.

It is reasonable to suggest that the development of high utility attri-
butes in person perception takes place through such a process. A low
utility attribute is directly tied to a particular stimulus person--it is
not recognized with regard to any other person, and only becomes salient
to the perceiver when the associated person is present or is remembered.
In contrast, a high utility attribute can be conceived of as a cognitive
structure through which many stimulus persons are construed. The per-

ceiver may make many inferences regarding a stimulus person having such

15
an attribute, since a large variety of other peOple, each with their own
associated attributes, are represented as "members" of the high utility
attribute.
Bruner, Olver, and Greenfield (1966) have conceptualized cogni-
tive development in a slightly different manner. For these researchers,
the development of the ability to form cognitive representations of

reality begins with enactive representation, in which a concept is a

 

bodily movement that transcends the serial linking of stimuli and

responses, is followed by ikonic representation, in which a concept is a

 

visual image of reality, and is completed with symbolic representation,

 

in which concepts are abstracted in the form of language. Despite the
elaborate nature of this formulation, it seems clear that the major
dimension of growth that is implied lies in the direction of increasing
distance between concept and physical reality. Enactive and ikonic
representations both require immediate involvement with a particular
stimulus; one cannot form a generalized representation when the medium
for this representation is physically or visually concrete. Symbolic
representation, on the other hand, allows abstraction from the immediate
environment, in that entire classes of stimulation may be grouped cog-
nitively within a single representation. In terms of person perception,
it can be inferred that a low utility attribute could conceivably be
represented in any of the three modes, while a high utility attribute
would most likely be represented symbolically. Again, it may be con-
cluded that the probability of occurrence of high utility attributes
increases with development.

A third line of evidence suggesting the developmental salience of
attribute utility comes from the work of the developmental psycholin-

guist, Lev Vygotsky (1962). To understand the significance of linguistic

l6

findings for building a model of person perception, it is necessary to
recognize the similarity between an attribute and a word. An attribute,
like a word, is a semantic unit whose meaning is acquired through experi-
ence by the individual. Many of the processes postulated to account for
the acquisition of 1anguage--a generalized symbolic system--should be
applicable to the acquisition of attributes--a specific symbolic system.
This supposition should be especially valid in the case of words which
resemble attributes. Nouns and adjectives which refer to physical
objects, for example, seem particularly similar to person attributes,
since both typescflfsymbols may be conceptualized as "categories" into
which objects of cognition may be grouped.

Keeping this analogy in mind, consider the statement of Vygotsky
(1962), that "There is every reason to suppose that the qualitative dis-
tinction between sensation and thought is the presence in the latter of a

generalized reflection of reality . . . [p. 5]." According to Vygotsky,

 

the child's major activity in development is the formation of generaliza-
tions, and words are the symbols for these generalizations. The Soviet
researcher undertook a detailed investigation of the development of con-
cept formation to demonstrate the generalizing function of words. The
very young child, he found, attempted to form a concept (a generaliza-
tion) by putting together a number of objects in an unorganized congerie--
a "heap." Later, children form complexes, groups formed on the basis of
factual, concrete bonds, such as "members of a family" or "things that
all belong in the kitchen." Further in development, children form gheip
complexes, groups in which the decisive attribute keeps changing during
the process of formation. The final developmental stage of concept for-
mation is typified by groups formed on the basis of single attributes

held in common by all members of the group. This developmental sequence

17
clearly implies that the use of high utility attributes occurs only late
in development. Earlier groupings, and their associated cognitive struc-
tures, are in a sense artificial, in that they are formed without respect
to any overall organizing principle, and are therefore of little use in
making generalized judgments about all members of the group-~or about
differences in "degree" of group membership.

Vygotsky argued on the basis of this analysis that the development
of word meaning follows a similar course, and it may be argued along the
same line that the development of attributes can be represented by these
stages. This parallel is particularly vivid in light of the similarity
between the idea presented earlier--that the "fixedness" of an attribute
in the cognitive Space is a function of the utility of the attribute-~and
an example of the development of a word. Vygotsky cites the example of
". . . a child's use of qpeh to designate first a duck swimming in a
pond, then any liquid, including the milk in his bottle; when he happens
to see a coin with an eagle on it, the coin is also called 3223’ and then
any round, coinlike object [1962, p. 70]." The word "quah" undergoes
endless changes, revealing the possibility recognized by Vygotsky, that
"one word may in different situations have different or even opposite
meanings . . . [p. 70]." From the perspective of the spatial model, the
constant shifting in "meanings” of words is caused by their newness to
the child--and their consequent low level of utility. For a person
attribute to have a stable meaning, it must be fixed in cognitive Space;
the child's collection of objects to which the word "quah” refers is
analogous to a collection of peOple to which a low utility attribute
refers. Only with repeated pairings of an attribute and various objects
does the "meaning" of the attribute become constant; moreover, the repe—

tition of such pairings is function of experience and deveIOpment.

18

What is being suggested, then, in view of the variety of cogni-
tive developmental evidence that has been summarized, is that the utility
of attributes employed in the construal of peOple increases with the
deve10pment of the perceiver. The theme of cognitive-developmental the-
ory, that the successive cognitive representations through which the
individual interprets the world become increasingly veridical and complex
as development progresses, is reflected quite clearly in this view of

person perception.

The Development of Person Perception:
Previous Research

 

 

Studies of the development of person perception have isolated a
number of age-related changes in both structural and content variables.
It has been shown, for example, that with development, descriptions of
people become less egocentric (HOpkins, Press, & Crockett, cited by
Olshan, 1970; Peevers & Secord, 1973), conflicting information about
people is better resolved (Biskin & Crano, 1973; Crockett, 1970; Gollin,
1958), and the major dimensions representing trait co-occurrence judg-
ments undergo changes (Olshan, 1970).

The unit of analysis of present concern--the attribute--also
appears to progress with development, as is evidenced by the results of
Signell (1966), who obtained a variety of measures of the structural
properties of interpersonal cognition on an age cross-sectional sample
(9-16) of perceivers. Two of the measures, tapping what she called the
"complexity of single concepts," are of Special interest, since they are
similar to the measures of attribute articulation employed by Bieri et
a1. (1966) and Scott (1969). The data on which these measures were
defined consisted of subjects' ratings of 24 different stimulus persons

along 7-point semantic differential scales representing judgmental

19
dimensions the subjects themselves had provided. The first measure of
articulation was the number of intervals of the 7-point scale the subject
used in rating the stimulus persons (acquaintances), while the second
measure was the information theory statistic, H (cf. Attneave, 1959; Gar-
ner, 1962), computed on each scale. The H statistic is a measure of the
"levelness" of the distribution of stimulus persons across intervals of
the scale, and can be thought of as an index of the scale's articulation,
Since "levelness" implies that the scale is used as a dimension rather
than a Simple dichotomous category. Signell found that both measures of
the articulation of attributes were highly correlated with the ages of
perceivers. It is interesting to note that of twelve structural measures
of person perception employed in this study, only these two (the sole
measures of attribute articulation) were significantly related to level
of development. This study does not, of course, provide any evidence on
the proposed relationship between utility and age, but instead indicates
that the hypothesized link between development, utility, and subsequent
articulation has some empirical support; attribute articulation in person
perception does increase with the age of the perceiver.

A second relevant finding seems, by itself, rather pedestrian:
the number of attributes given by the perceiver in free descriptions
increases with the age of the perceiver (Biskin & Crano, 1973; Livesley &
Bromley, 1973; Yarrow & Campbell, 1963). It is obvious that age-related
increases in verbal skills alone might account for this finding--although
a number of researchers have argued that the number of attributes is a
reasonable intuitive measure of the perceiver's level of differentiation
(cf. Crockett, 1965, 1970; Livesley & Bromley, 1973). This label will be
adopted in the present discussion, not because of any theoretical commit-

ment, but because it has been commonly used by previous writers. What is

20
important to recognize, however, is that level of differentiation is per-
tinent to the measurement of the individual perceiver's "overall level of
attribute utility," since such an overall utility measure can only be
assessed in relation to the total attribute population employed by the
perceiver; it is insufficient, for example, to prOpose that the absolute
number of high utility attributes increases with development, when this
number may simply be a constant fraction of the increasing total number

of attributes employed. The proportion of high utility attributes in the

 

perceiver's total set of attributes is a more meaningful measure of the
"average” utility of the perceiver's cognitive structure. Scott (in
press) and Kelling (1968), in defining measures of "image comparability,"
an individual difference variable similar in many respects to an "aver-
age" utility measure, recognized this methodological difficulty and
resolved it in a similar manner, treating utility as a proportion of

attributes instead of an absolute number.

Summary and Hypotheses

An informal model of interpersonal cognitive structure, the theo-
retical perspective of cognitive develOpment, and evidence from develop-
mental studies of person perception have all been discussed in the service
of formulating hypotheses regarding the development of "core categories"
employed by perceivers in construing others. The sum of the arguments is
that the utility of attributes--the number of stimulus persons to which
they refer-~18 positively related to the perceiver's level of development,
and in addition, is positively related to the articulation of the attri-
butes themselves. This proposition was investigated in two experiments,
the first of which examined the develOpmental segment. One hypothesis

generated for this experiment was offered for the purpose of the

21
replication and extension of previous developmental studies, specifically
those which have reported a developmental increase in differentiation
(Biskin & Crano, 1973; Livesley & Bromley, 1973).

Hypothesis 1. The total number of different attributes given by

subjects in free descriptions of acquaintances will be greater

for older than younger subjects.
The other result expected in the developmental study was that attribute
utility would increase with level of development. The statement of this
hypothesis reflects the methodological considerations outlined earlier.

Hypothesis 2. The ratio of the number of high utility attributes

to the total number of different attributes given by subjects in

the free description of acquaintances will be greater for older
than younger subjects.

The second experiment was designed to examine directly the effect
of utility upon the articulation of attributes. The hypothesis associated
with this experiment, derived from the model elaborated earlier, is cen-
tral to the explication of the developmental function of attribute
utility.

Hypothesis 3. The articulation of subjects' judgments of their

acquaintances will be greater for high than low utility attributes.

EXPERIMENT l

The principle aim of this study was to test Hypothesis 2, regard-
ing the developmental aSpects of attribute utility. The experiment was
designed, in addition, to provide information on the development of dif-
ferentiation (Hypothesis 1), and on the possible influences of sex and
verbal intelligence upon both utility and differentiation. Since both
individual differences in verbal intelligence and sex differences in ver-
bal fluency were expected to increase the variability in differentiation
among subjects (cf. Biskin & Crano, 1973; Crockett, 1965, 1970), the two
variables were included in the analysis to offset potential problems in

examining the relationship between differentiation and development.

Method
Subjects
A cross-sectional sample of 88 students drawn from the fourth,

sixth, ninth, and twelfth grades of the Holt, Michigan public school sys-
tem served as subjects. Approximately equal numbers of males and females
comprised each grade level group; the exact sample sizes and mean ages of
each sex by grade group are diSplayed in Table 1. Grade was retained as
the classificatory variable for age since there was no overlap between age

ranges associated with grades.

Free Description Procedure
A number of researchers have found that subjects' free response

descriptions of others can be reliably coded into systems of content

22

23
TABLE 1

Sample Sizes and Mean Ages (Years, Months) by Grade and Sex

 

 

 

 

 

Grade
Sex
4 6 9 12
N 10 9 10 14
Males
Age 10,0 12,3 15,1 18,3
N 10 11 12 12
Females
Age 10,0 12,0 15,1 18,2

 

 

 

 

 

categories (see, e.g., Beach & Wertheimer, 1961; Dornbusch et al., 1965;
Faguy-Coté, 1965; Livesley & Bromley, 1973; Yarrow & Campbell, 1963).
The advantage of such a procedure over the more typical data collection
strategies (adjective checklists, rating scales, and so on) is that the
free description data is more natural-~subjects use their own "person
concepts" in their own way (cf. Peevers & Secord, 1973). The disadvan-
tage, of course, is that the coding of free descriptions requires judges
to make a large number of decisions about a subject's responses, many of
which may misconstrue the subject's intended meanings. This response
format, however, seemed clearly desirable for the present study, since
the wide variation in subjects' ages was a factor which would needlessly
limit the attribute vocabulary used in a more structured instrument.

For the administration of the free description task, subjects
were assembled in groups of 3-5 (all members of the same grade) in a
testing room in their school building. After being informed that their

responses would be completely confidential ("secret" for the fourth and

24
sixth graders), and would be used "to find out what students, in general,
think about the peOple they know," subjects were introduced to the expe-
rimental task. Each subject was initially asked to write down the names
of eight acquaintances in reSponse to role descriptions similar to those
employed by Crockett (1965). The role descriptions consisted of state-
ments of age (peer or older person), sex (male or female), and liking
(liked person, disliked person), Such that each of the eight stimulus per-
sons was specified by a different combination of the three variables.
Thus two of the descriptions were, for example, "a boy near your age you
dislike," and "an older woman you like.”

Care was taken to make the nature of the free description task
clear to all subjects. Ninth and twelfth grade subjects received writ-
ten instructions which were read aloud to them by the experimenter;
fourth and sixth grade subjects had no written instructions, but rather
heard the experimenter explain the task twice. Subjects were asked to
write (or print) "what you think about each person" on a separate page in
the teSt booklet. The instructions, designed to elicit a full range of
responses (cf. Livesley & Bromley, 1973), also included the suggestion
that subjects try to answer the question, "What is this person like?"

The order of appearance of stimulus persons in the test booklet was sys-
tematically varied among subjects. Although there was no time limit
imposed on the description task, all subjects completed it in less than

one and one-fourth hours.

Coding System

 

Since a number of coding systems have been designed for the anal-
ysis of free descriptions of people into the categories of description

(cf. Dornbusch et al., 1965; Faguy-Coté, 1965; Livesley & Bromley, 1973),

25
it was possible to create the present system as a reflection of both pre-
vious systems and the present data. Ten undergraduate students (who were
unaware of the hypotheses of this research, and who later served as
coders) were familiarized with previous systems and were asked to help
devise a system suitable for the present analysis. They were first
trained in identifying an "attribute," under the instructions that an
attribute consists of "a unit of meaning--a word, phrase, or sentence--
referring to the stimulus person, which in the context of the entire
description is not divisible into smaller units."

Based on practice with applying this definition, the coders
derived three preliminary "approximations" to the final system by identi-
fying frequently used attributes in the present data, testing their effi-
ciency in summarizing the population of attributes, and subsequently
deleting or adding new categories. The final system (see Appendix A) was
useful in categorizing 81% of subjects' responses, either as "true attri-
butes" (statements similar in meaning) or as "attribute classes" (head-
ings which served to summarize the data--"occupation," for example, was

' and ”mailman"--

used to classify attributes such as "teacher," "lawyer,'
but which were not treated as single attributes in the derivation of
dependent measures). The remaining 19% of the attribute population, usu-
ally specific examples or little-used terms, were nevertheless included
in the final analysis. In essence, then, the coding system was a method
of consolidating a vast quantity of data; those attributes not summarized
within the system were treated separately in the derivation of measures,
but in a manner consistent with that used for coded attributes. The
final system of 51 categories (47 single attributes and 4 attribute

classes) was supplemented by further specification of positive and nega-

tive instances within each attribute (e.g., "helpful" and "not helpful").

26

Coding Procedure and Intercoder Agreement

 

The 10 coders who designed the system were not given individual
access to all of the data during the design process, thus each was
unfamiliar with the particular set of booklets he received for final cod-
ing. To assess intercoder agreement, every subject's protocol was coded
twice, once by each of two coders.

Two types of intercoder agreement were assessed-~the agreement on
the "definition of an attribute," and the agreement on the subsequent cod-
ing of attributes into the category system. The measure of agreement on
attribute definition, for a particular subject's booklet, was the number
of identical attributes isolated by the two coders divided by the average
total number of attributes identified in the protocol by the two coders.
Expressed as a percentage, this index of agreement on a single protocol
ranged from 93% to 100%, with a mean value across protocols of 96%. The
second index of agreement, for a particular subject's booklet, was the
number of identical attributes isolated by the two coders which were also
identified as members of the same coding category, divided by the average
total number of attributes identified in the protocol by the two coders.
Although this second index is obviously dependent upon fluctuations in
the first index, the range of values for subjects' booklets was satisfac-
tory (77%-98%), and the mean value across subjects (82%) was also high.
In sum, the intercoder agreement obtained in the analysis of the data was
considered satisfactory with regard to both attribute definition and

categorization.

Design and Measures

 

Hypothesis 1, that the total number of different attributes given

by subjects describing people increases with the subjects' age, was

27
examined using a 2.x 4 factorial design with sex and grade of subject as

the independent variables. The dependent measure (differentiation score)

 

was the number of different attributes used by the subject in describing
all eight stimulus persons. This number included not only the number of
attribute categories to which codable attributes were assigned, but also
the number of additional different uncodable attributes. In this second
case, any two attributes were judged as similar only if the same word or
words appeared in both (negations were considered similar); all other
attributes were classified as different.

Hypothesis 2, that the utility of attributes given by subjects
describing people increases with the subjects' age, was also examined
using a 2 x 4 factorial design with sex and grade of subject as indepen-

dent variables. The dependent measure (utility ratio) was defined as the

 

number of high utility attributes used by a subject in describing the
eight stimulus persons, divided by the total number of different attri-
butes used in describing the eight persons. A high utility attribute was

defined as an attribute the subject used to describe more than one of the

 

eight persons. The utility ratio, whose numerator is the total number of
such attributes, and whose denominator is the subject's differentiation
score, could conceivably vary between 0 and 1. A zero utility ratio
would indicate that no attribute was used to describe more than one Stim-
ulus person, while a ratio of one would indicate that every attribute
mentioned was used in describing more than one stimulus person. AS in
the calculation of the differentiation score, the utility ratio was based
on both codable and uncodable attributes. For codable attributes, the
use of a category in describing more than one person defined the category
as a high utility attribute; for uncodable attributes, similar attributes

used in describing more than one person were defined as high utility

28
attributes. In this second case, judgments of similarity followed the
same criterion used in calculating differentiation scores, i.e., similar
attributes are those stated in the same word or words, with negations
considered similar to their positive instances.

Although the expected relationships between the dependent mea—
sures and the independent variable of age were of prime theoretical
interest, the effect of sex was also assessed, Since sex differences in
verbal fluency (cf. Terman & Tyler, 1954) were expected to increase the
variability of both differentiation scores and utility ratios within each
grade level group. In addition, the effects of individual differences in
verbal intelligence were of interest, Since such differences could also
increase the variability of dependent measures within age groups. Verbal
IQ scores on the school-administered Lorge-Thorndike Intelligence Test
(Lorge & Thorndike, 1954) were obtained for all subjects. Fourth and
sixth grade subjects had been tested during their fourth grade year,
while ninth and twelfth graders had taken the test in their seventh grade
year. In sum, the inclusion in the experimental design of sex as an
independent variable and verbal IQ as a potential covariate was planned
for the purpose of increasing the salience of age (grade) as an indepen—

dent variable.

Results

PreliminaryjAnalyses

 

To investigate the possible biasing influence of verbal intelli-
gence upon the dependent measures, an initial correlational analysis was
undertaken. It was found that differentiation scores were significantly
correlated with verbal IQ (g = .39, p_< .001), whereas utility ratios

were not (E = .02). (The means, standard deviations, and matrix of

.‘Dr rail:

29
correlations among age, verbal intelligence, differentiation scores, num-
ber of high utility attributes, and utility ratios are shown in Appendix
B). In accord with this finding, verbal intelligence was treated as a
covariate in the subsequent analysis of differentiation scores, but was

not included in the analysis of utility ratios.

Differentiation

 

A 2‘5 4 factorial unweighted means analysis of covariance (cf.
Winer, 1971, p. 792) with Sex and Grade of subject as the independent
variables was employed. Subjects' differentiation scores served as the
criterion measure, and verbal intelligence scores served as the covariate.
Table 2 shows both the original and adjusted cell means for the differen-

tiation measure.

TABLE 2

Original and Adjusted Cell Differentiation
Means by Grade and Sex

 

 

 

 

 

Grade
Sex
4 6 9 12

Original 22.1 29.4 25.2 29.2
Males

Adjusted 22.0 29.6 27.0 28.4

Original 22.2 34.0 35.3 40.2
Females

Adjusted 22.8 33.1 36.3 38.2

 

 

 

 

 

The analysis of covariance revealed significant main effects for Grade

F = 6.04, 531: = 3/79, 2 < .005) and for Sex (I: = 8.49, g_f_ = 1/79,

p’< .005); the Grade by Sex interaction was not significant. (Summary

30
tables for the analysis of variance and the analysis of covariance are
Shown in Appendix C).

A decomposition of the significant main effect for Grade (Newman-
Keuls procedure) revealed that the fourh grade Subjects' mean differentia-
tion scores (M = 22.4) were significantly lower than those of the sixth,
ninth, and twelfth grade subjects (M = 31.4, 31.7, and 33.2, respectively;
p_< .01 in each case). None of the other possible comparisons of differ-
entiation means among grade levels reached statistical significance. It
is interesting to note, however, that the differentiation means increased
monotonically with grade level (22.4, 31.4, 31.7, and 33.2, from fourth to
twelfth grades, respectively). Hypothesis 1 was generally confirmed,
since these findings provide support for age-related increases in differ-
entiation scores, especially between the ages of 10 and 12 years (fourth
and sixth grade levels).

The direction of the Significant main effect for Sex was toward
greater differentiation by females (M = 32.6) than by males (M = 26.8).

An inspection of the adjusted cell means in Table 1 Shows that this
effect was evident at every grade level. Since potential sex differences
in verbal intelligence were removed from the data, it would seem that
this result is not interpretable in terms of sex differences in verbal
skills.

In sum, the results substantially support the notion that differ-
entiation increases with level of development, and the results also sug-
gest that females are more differentiated than males at all age levels
studied. It should be noted that an analysis of variance in which verbal
intelligence was not a covariate produced comparable results, both for

sex and age effects.

31

Utility Ratio

 

Since the preliminary analysis indicated that verbal intelligence
was not related to the utility ratio measure, IQ was not treated as a
covariate in the analysis. A 2‘5 4 factorial unweighted means analysis
of variance with Sex and Grade of subject as the independent variables
was employed. Table 3 shows the utility ratio cell means by grade and

sex. The analysis of variance revealed a Significant main effect for

TABLE 3

Utility Ratio Cell Means by Grade and Sex

 

 

 

 

 

Grade
Sex
4 6 9 12
Males .153 .149 .191 .283
Females .179 .153 .215 .292

 

 

 

 

 

Grade (E = 13.23, gf = 3/80, p.< .001); the main effect for Sex was not
significant, nor was the interaction effect. (A summary table of the
analysis of variance appears in Appendix D). Subsequent multiple compari-
sons among utility ratio means for all grade levels (Newman-Keuls proce-
dure) showed that the twelfth grade Subjects had Significantly higher
utility ratios (M = .287) than did those in the ninth (M_= .204,

p.< .01), sixth (M = .151, E.< .01), or fourth (M = .166, p_< .01) grades.
In addition, the ninth grade mean was significantly higher than the sixth
grade mean (M = .204 versus .151, respectively; p|< .05). No other dif-

ferences among grades reached significance.

32
Although the progression in utility ratios with higher grade
levels was not strictly monotonic (.166, .151, .204, and .287), the anom-
alous pair of means (for fourth and sixth grades) were not significantly
different; the Significant differences encountered were all in the direc-
tion of increasing utility.with age. The pattern of results is highly
supportive of Hypothesis 2, in that the proportion of high utility attri-
butes to total number of different attributes given by subjects in their
free descriptions of eight acquaintances--the utility ratio--was found to

increase significantly with the age of the subject.

Summary

Both measures of interpersonal cognitive Structure, differentia-
tion scores and utility ratios, were found to increase with the age of
the perceiver. For differentiation scores, the major portion of this
change occurred between the ages of 10 and 12; for utility ratios, the
major change occurred between the ages of 15 and 18. While sex differ-
ences in differentiation were observed, with females being more differen-
tiated than males, no such differences were found in utility ratios.
Individual differences in verbal intelligence were positively related to

differentiation, but were unrelated to utility ratio.

EXPERIMENT 2

This study was designed to test Hypothesis 3, regarding the rela-
tionship between attribute utility and attribute articulation. While the
results of Experiment 1, which found significant age-related increases in
the proportion of high utility attributes used by perceivers, and the
results of Signell (1966), who found age-related increases in mean attri-
bute articulation, when taken together are seemingly indicative of such a
relationship, they are not in any clear sense supportive of a connection
between the utility and articulation of an individual attribute for a par-
ticular perceiver. Similarly, the evidence provided by Scott (in press),
that individual differences in articulation and utility ("image compar-
ability") are positively related, is only minimally applicable to Hypo-
thesis 3. The present experiment was designed, therefore, to investigate
directly the relationship between the utility and the articulation of the

individual attributes used by individual perceivers.

Method

Subjects and Design

 

Twenty undergraduates (10 males and 10 females) enrolled in
introductory psychology classes at Michigan State University agreed to
take part in the study in exchange for extra course credit. Hypothesis
3, that the articulation of subjects' judgments of their acquaintances
is greater for high than low utility attributes, was the focus of this

study. The effect of attribute utility upon attribute articulation was

33

34

assessed by measuring each subject's mean articulation for a sample of
his high utility attributes and for a sample of his low utility attri-

butes. Two independent measures of articulation were employed.

Attribute Selection Procedure

 

To select a sample of the high utility and the low utility attri-
butes employed by each subject, the free description task outlined in
Experiment 1 was administered; the testing sessions were attended by
groups of three to five subjects. The subsequent coding of these proto-
cols by two coders also followed the procedure of the earlier experiment.
The mean intercoder agreement was 99% for attribute definition and 84%
forcategorization. High utility attributes and low utility attributes
were identified (from both the codable population and the uncodable p0pu-
lation) for each subject. A high utility attribute was defined, as in
Experiment 1, as an attribute used in describing more than one of the
eight stimulus persons, and a low utility attribute was, in turn, an
attribute used to describe only one of the stimulus persons.l From the
two samples of attributes obtained from each subject, a set of five high

utility and a set of five low utility attributes were randomly selected.

Articulation Measurement Procedure

 

Each subject returned for individual testing within three days of
the group session. Both articulation measurement procedures required
that the subject make judgments of stimulus persons (the eight acquaint-
ances originally identified for the free deScription task) on each of the
10 attributes selected from his protocol. To facilitate this task, the
subject was first asked to examine each attribute and to create anchors
for a bipolar dimension based on the attribute. The experimenter pro-

vided the following example of this procedure:

35

Suppose that you had written a description of an oak tree, and had

said that it was "fairly tall." Now if I wanted you to rate a vari-

ety of trees on how ”fairly tall" they were, it might be easier for

you to do if you thought of a dimension of "very short" to "very

tall."
The subject was told that to help along with this procedure, the experi-
menter would ask him to "tell me the Opposite" of an attribute; this pro-
vided one endpoint of the dimension. To elicit the other endpoint, the
subject was told that the experimenter would say "now, tell me the oppo-
site of that." Subjects readily understood this procedure; the eXperi-
menter then began the task by presenting to the subject one of the 10
attributes, and eliciting from the subject his dimensional anchors. The
10 attributes were presented to the subject in a random sequence.

The order of the two tasks the subject was subsequently asked to

perform, the two articulation measures, was systematically varied across

the subjects. One articulation measure (number of groups) was a measure

 

defined by Scott (1969), in which the subject is asked to sort stimulus
persons into groups on the basis of a dimension. For this measure, the
subject was given cards with the names of each of the eight acquaintances
originally identified for the free description task. Returning to the
"tree tallness" analogy, the experimenter gave the following example of
the card sorting task:
Suppose you were to sort redwood, elm, maple, and peach trees accord-
ing to the "very short--very tall" dimension. Of course, a redwood
would be nearest "very tall," and a peach tree might be the Shortest,
so you would put it nearer the "very Short” end. But, say you knew
an elm and a maple were between the two extremes, but you couldn't
tell which was usually the taller of the two. Well, then you could
put them in the same pile, in the middle of the dimension.
Every subject readily understood this procedure; groupings for each
dimension (defined by the subject's own anchors) were recorded. The

index of articulation for each of the high and low utility attributes, the

number of groups formed by the subject on the basis of that attribute

. ‘55... .Iv. : \airfiulIng

a

36
dimension, could vary between 1 and 8, with higher scores indicating
greater articulation.

The other measure of articulation was designed specifically for
the present study on the basis of similar measures originally devised by
Ulehla (1961) and Signell (1966). In particular, the subject's task was
to "rate" each of the eight stimulus persons by writing the number cor-
reSponding to each in one of a set of five "boxes." Squares arranged
horizontally on the test sheet were used to represent an attribute dimen-
sion, with the end squares labelled by the anchors the subject had pro—
vided. The instructions stressed the "rating" nature of the task, and
also pointed out that boxes could be left empty, filled with all eight
persons, or used to any intermediate extent. The subject completed this
task for each of the selected high and low utility attributes.

The M_statistic, a measure derived from information theory (cf.
Attneave, 1959; Garner, 1962), was calculated for each of the subject's
attributes. If the proportion of the eight stimulus persons in the ith
interval of the 5 interval dimension is Pi’ then '

P1 1082 P1
1

In:
II
I
u Flux

M.is the number of "bits” of information contained in the distribution of
stimulus persons on the attribute dimension. It can be thought of as a
measure of the "levelness" of the distribution, or, alternately, as a
measure of variance which is independent of the central tendency of the
distribution (cf. Coombs, Dawes, & Tversky, 1970). This statistic, a
summary of the subject's sorting of eight stimulus persons into a maximum
of five scale intervals, could conceivably vary between 0 (in the case of
all eight stimulus persons grouped in the same interval) and 2.25 (in the

case of maximum distribution of stimulus persons, i.e., any sorting of

37

the variety 2-2-2-1-1, regardless of order), with higher values indicat-

ing greater articulation of the attribute.

Results

Relationship Between
Articulation Measures

 

 

An initial analysis was performed to explore the level of rela-
tionship between the two measures of attribute articulation. One index
of this relation was calculated between the two measures at the level of
individual differences among subjects; the other index revealed the rela-

tionship at the level of single attributes within subjects.

 

For the subjects measure, a mean score on the number of groups
articulation measure and a mean score on the information articulation
measure were calculated for each subject across his 10 selected attri-
butes. The correlation of these two means across subjects was not sta-
tistically significant (3 = .26, M_= 20, £.> .10). Although this
correlation is of the same order of magnitude as those reported by Scott
(1969, in press) for similar individual difference measures of articula-
tion, it is not indicative of any relationship between the present mea-
sures at the individual difference level.

For the attributes measure, a correlation for each subject was

 

computed between the number of groups score and the information scores
across the subject's 10 attributes. This index ranged, among subjects,
from E = .19 to‘g = .86. The mean value for all subjects (f = .52) was
tested against the null hypothesis of‘f = 0, and was found to be highly
significant (5 = 12.14, g: = 19, p.< .0001). It was concluded that at
the level of the individual attribute, within each subject, the two mea-
sures of attribute articulation were closely related. The pattern of

these two measures of relationship suggests that analysis at the level of

'5... n... L: .5...1J.§1._
b

38

individual differences grossly compromises the sensitivity of articula-
tion measures to the differences among individual attributes employed by

the single perceiver.

The Effects of Attribute Utility

For each subject, means on the two measures of articulation, num-
ber of groups and information, were computed separately for the five high
utility attributes and the five low utility attributes. The within sub-
ject means were then averaged across subjects; these mean articulation
Scores (number of groups and information) for high and for low utility

attributes are shown in Table 4.

TABLE 4

Means of Within-Subject Means on Number of Groups
and Information Measures of Attribute
Articulation for High Utility
and Low Utility Attributes

 

 

Attribute Utility

Articulation Measure

 

 

 

Low High
Number of groups 3.80 4.48
Information (M) 1.58 1.74

 

 

 

For the number of groups articulation measure, a.£ test for
related measures performed on the mean of within-subject means for high
utility (5.: 4.48) and for low utility (M = 3.80) attributes revealed a
significant difference between means (5 = 5.51, ME = 19, p.< .0001);
Subjects formed more groups of the eight stimulus persons on the basis of

their high utility than their low utility attributes. This effect was

39
evident for all 20 subjects--no reversals of the direction of the differ-
ence between means were observed.

The significance of the effect of attribute utility upon the
information measure of articulation was similarly assessed with a.£ test
for related measures. The mean of within-subject information means for
high utility attributes (M = 1.74) was significantly greater than that
for low utility attributes (M = 1.58; E = 3.63, g: = 19, p'< .01). Dif-
ferences in the high and low utility means were in this direction for 17
of the 20 subjects.

Hypothesis 3 was generally confirmed, in that subjects displayed
greater articulation in dimensional judgments of eight acquaintances
based on their selected high utility attributes than on their selected
low utility attributes. This result was obtained using both the number
of groups measure and the information measure of attribute articulation.
The two articulation measures, although not significantly related as
individual differences among subjects, were highly related at the level

of the individual attribute within subjects.

DISCUSSION

The results of Experiments 1 and 2, viewed as a whole, generally
verify the develOpmental and cognitive structural significance of attri-
bute utility. The specific results are taken up as they relate to the

discussion of each of the hypotheses.

Differentiation and Development

 

Hypothesis 1, that the number of different attributes given by
perceivers describing stimulus persons increases with the age of the per-
ceiver, received substantial support in the first experiment. Congruent
with the earlier findings of Biskin and Crano (1973), Livesley and Bromley
(1973), and Yarrow and Campbell (1963), it was found that differentiation
increased with the age of the perceiver. It is interesting to note that
the particular age trend observed in the present sample, with the great-
est increase in differentiation occurring between the 10 and 12 year old
groups, has not been previously reported. It should be emphasized that
the general age-related increase in differentiation obtained in the pres-
ent study seems to be a reliable and replicable effect, but that the Spe-
cific (10-12 year old) age trend observed was not predicted in advance
and should be interpreted only with caution.

The general age-related increase in differentiation was found to
hold even when the effects of verbal intelligence were eliminated. Since
intelligence scores are usually represented as "mental age" divided by

chronological age (IQ = MA/CA), the analysis of covariance had the

40

41

function of "homogenizing" the verbal ability of each grade level group

according to their chronological age. Thus, although this methodological
process reduced the variation due to verbal ability to that associated
with differences in chronological age, it did not serve to rule out the
possibility that developmental increases in the number of different
attributes employed by perceivers are Simply a function of age-related
increases in verbal ability or verbal fluency.

The highly significant effect of sex upon level of differentia-
tion, with females diSplaying greater differentiation than males, is not
interpretable in terms of sex differences in verbal intelligence, since
such differences were covaried out of the analysis. Three tentative
explanations of the sex effect may be offered. Taking the result at face
value, it could be suggested that females recognize more different attri-
butes of stimulus persons than do males because their interpersonal cog-
nitive structures are actually more differentiated. This explanation is
supported by the general finding that females have greater ”social sensi-
tivity" than males (Maccoby, 1966); greater awareness of social stimuli
on the part of females has been observed in a variety of empirical set-
tings. A second line of reasoning regarding the sex difference in dif-
ferentiation is that sex differences in verbal abilities may not have
been sufficiently controlled. Measures of verbal intelligence may not
tap differences in verbal fluency, and since females are more verbally
fluent than males (cf. Terman & Tyler, 1954), the sex difference in dif-
ferentiation might have been a function of the amount of verbal output
elicited from each sex group. The third plausible explanation of the sex
effect is that subject task motivation may have been a differential func-
tion of sex; this eXplanation is especially credible since a male experi-

menter administered,the free description task to most of the subjectS—-a

{Hula

I If: . II: .I». [I ..s.‘!‘|¥iJ
Irlf u. .x

42
female experimenter was used in only 15% of the testing sessions. This
factor might have encouraged female subjects to become more involved in
the task. Given the variety of competing explanations applicable to the
observed sex effect, and the equivocal evidence provided by previous
research (Biskin & Crano, 1973; Crockett, 1965; Livesley & Bromley,
1973), the present finding of greater differentiation in females than

males must be viewed as merely suggestive of such a relationship.

Attribute Utility and Development

Hypothesis 2, that the "average” utility of attributes given by
perceivers increases with the developmental level of the perceiver, was
substantially confirmed in Experiment 1. This effect was especially pro-
nounced in the comparison between 15- and 18-year-old subjects--the great-
est increase in utility ratio occurred between these age groups.

Although this particular result was not anticipated, it is suggested that
future research extend the measurement of attribute utility to even older
groups; such research might indicate more completely the shape of the
"age--utility" curve.

The general finding of increased attribute utility with age might
be considered less questionable than the finding of increased differen-
tiation, Since problems in the interpretation of the differentiation
result arising from the observed relationship between differentiation and
verbal intelligence are not applicable to the utility result; utility
ratio was not related to verbal intelligence. A more Specific statement
of this finding (based on the IQ = MA/CA formula) is that utility ratios
were not related to differences in mental age within chronological age
groups (grades). Yet this interpretation oftimelack of correlation

between IQ and utility ratio must be viewed with circumSpection; since

43
the utility ratio is a ratio of number of high utility attributes to
total number of attributes, it is conceivable that the variance due to
verbal intelligence in both the numerator and denominator of the ratio
was removed by the division process, and was therefore absent in the quo-
tient. Because of this methodological difficulty, the question of the
relationship between "average" utility and verbal intelligence must
remain until a suitable measure of "average" utility (i.e., one not for-
mulated as a ratio) is devised.

Fortunately, the exact specification of the relationship between
utility and verbal intelligence is not central to an understanding of
utility. Unlike differentiation, a measure which logically is dependent
upon the verbal skills of the perceiver, the utility ratio is not con-
founded by such a connection. Thus, while the empirical independence of
utility ratio and verbal intelligence cannot be examined in the present
study (because of the difficulties outlined previously), the logical inde-
pendence of the two variables seems obvious. At the lowest limit, of
course, a person with very low verbal intelligence--who is unable to ver-
balize any attributes at all--would have no high_utility attributes.
However, this example is extremely atypical; in the main, the proportion
of high utility attributes to total number of attributes given by a per-
ceiver is independent of the verbal skills of the perceiver. In sum, the
confounding influence of verbal intelligence (and verbal fluency), which
rendered problematic the interpretation of the observed relation between
differentiation and develOpment, does not impinge upon the interpretation
of the observed relation between utility ratio and development.

The general finding of increased utility of attributes employed
by perceivers at higher levels of development provides a bridge between

the two areas of developmental research discussed earlier-—cognitive

44

development and person perception develOpment. The frequent finding of
increased "generality" of concepts and words evident in the cognitive—
developmental studies of Piaget (cf. Flavell, 1963; Wohlwill, 1962),
Bruner et a1. (1966), and Vygotsky (1962) is directly applicable to the
process of person perception development. Given that attributes are the
"concepts" and "words" of interpersonal cognitive structure, their matu-
ration into the "core categories" of adult interpersonal thought through
increasing utility is reminiscent of the parallel ongoing processes of
word and concept generalization. In this sense, the development of per-
son perception and the development of object perception are isomorphic.

It might be reasonably concluded from the present result that the
final product of person perception development, a set of "core catego-
ries" used in the construal of others, is the proper subject for study in
future investigations of adult person perception. Although the present
research was not concerned with the content of these categories or with
their interrelations, future research would do well to investigate these
topics. Previous studies of adult person perception, often examining the
content and interrelations of unselected sets of attributes, have been
needlessly hampered by error stemming from the inclusion of low utility
attributes. When a subject is presented with the task of rating a stimu-
lus person on a low utility attribute-~one on which the subject has never
previously construed the person--he is actually being asked "to come up
with a cognition on the spot." Such judgments may be extremely variable,
since the perceiver is ”guessing" rather than "reporting." What is being
suggested, then, is that person perception research take a purely phe-
nomenological viewpoint--and not assume, as has commonly been assumed,
that any and every attribute presented to a subject will produce meaning-

ful judgments of stimulus persons. Some researchers have previously

..l)»»

'5... . é . was»; .I.’ a

ulus
k

45
recognized this necessity (Hastorf et al., 1958; Warr & Knapper, 1968),
and have argued the point out of a concern for the phenomenological
validity of empirical findings. In light of the present finding of
increased attribute utility with age, it should be noted that this admo—
nition must be most seriously considered in studying person perception in
children; but Since even adults seldom diSplay large proportions of high
utility attributes, care must also be taken in adult studies to provide

subjects with meaningful tasks.

The Utility and Articulation of Attributes

The single prediction associated with the second experiment,
stated in Hypothesis 3, was confirmed; perceivers were found to be better
able to make articulated judgments of their acquaintances on the basis of
high utility attributes than low utility attributes. One measure of
attribute articulation found to vary with attribute utility was the num—
ber of groups of stimulus persons formed by the perceiver with respect to
an attribute dimension; more groups were formed for high than for low
utility attributes, indicating that perceivers made finer discriminations
among the eight stimulus persons when they judged the persons according
to high utility attributes. The other measure of articulation was the
amount of information (M) associated with perceivers' ratings of the
eight stimulus persons on an attribute dimension; high utility attribute
judgments contained more information than did low utility attribute judg-
ments. This finding indicates that perceivers made more articulated
judgments of stimulus persons, "spreading out" the persons along the
attribute dimension rather than dichotomizing the persons or rating all
the persons as the same, when the attribute dimension had higher utility.

Given the admittedly coarse method of determining the utility of each

46
perceiver's attributes--a method based on only a meager sample of each
perceiver's attributes, and dependent upon responses to only eight stimu-
lus persons--the strength of the relationship between utility and articu-
lation evident for both articulation measures is certainly encouraging.

The relationship between the two measures of articulation, in a
sense an assessment of the reliability of attribute articulation measure-
ment, was found to be not significant at the level of individual differ-
ences in mean articulation across subjects, but to be highly significant
at the level of differences in articulation of individual attributes
within each subject. Since the actual assessment of the effect of attri-
bute utility upon articulation was based on differences in the articula-
tion of sets of high and low utility attributes within each subject, and
the effect of individual differences in mean articulation was eliminated,
it may be concluded that the general finding of a positive relationship
between attribute utility and articulation was based upon closely related
measures of articulation.

The observed relationship between the utility and articulation of
attributes is congruent with the prediction based on the informal model
of interpersonal cognitive Structure presented earlier. Indeed, such a
structural model seems necessary to represent the result, Since an analy-
sis of the actual content of an attribute is not sufficient to allow pre-
diction of such a finding. Consider, for example, the case of one subject
in the experiment who provided, among others, the attributes ”Michigan
State University Student” and "intelligent." The subject employed the
"MSU student" attribute with high utility, applying it to three of the
eight stimulus persons, but only mentioned the "intelligence" of one
stimulus person. Later, when the subject was asked to form groups of

stimulus persons according to each attribute (for the number of groups

47
measure of articulation), the subject made strikingly different groupings.
For the "MSU student--not an MSU student" dimension, the subject formed
four groups--which upon later questioning, he revealed to be MSU alumni,
full-time students, a student on academic probation, and nonstudents. In
contrast, the subject Simply dichotomized the Stimulus persons on "intel-
ligent--not intelligent," forming groups of five and three persons,
respectively. Clearly, a prediction of the differential articulation of
the two attributes based on the attribute content would require "MSU stu-
dent" to be less articulated than "intelligence," since the former seems
to be a dichotomous category and the latter a dimension amenable to fine
discrimination. Yet in this case, observation of the utility of each
attribute leads to a correct prediction of their differential articula-
tion. Since the significance of the relationship between the utility and
articulation of attributes was demonstrated in Experiment 2, it seems
reasonable to attempt further extension of the informal model responsible
for predicting the relationship.

One method of extending the usefulness of the model would be to
specify the actual parameters involved, with the goal of predicting more
(precisely the articulation of attributes from measures of their utility.
Future research might, for example, introduce novel attributes to sub-
jects, pairing each of the attributes with a different number of stimulus
persons; the utility of attributes could be manipulated experimentally.
It would be predicted, then, that subjects' judgments of stimulus persons
would be more articulated for higher utility attributes, and that the
functional relationship between utility and articulation might be quantita-
tively Specified. Future investigations might also dwell on the opposite

question: Does training in the articulation of a particular attribute

r bull? I. Eli. .wNI‘

48
increase the likelihood that that attribute will be applied to a greater
number of Stimulus persons? Since the present research was not addressed
to the direction of causality in the relationship between utility and

articulation, such an investigation would be of particular interest.

Conclusions

 

The deve10pment of person perception, as portrayed here, is a
process of change in the Structure of cognitive representations used in
the construal of persons. One such change is Simply the proliferation of
attributes with deve10pment; the young child has few attributes upon
which to base responses to stimulus persons, while the older child iden-
tifies many attributes. This increase in differentiation has the func-
tion of increasing the number of ways in which the perceiver may isolate
a stimulus person from the field of persons in general.

A more profound developmental change takes place in the structure
of the attributes themselves. With growth, some attributes increase in
utility-ethe perceiver employs them to represent more than one stimulus
person. As certain attributes are used with increasing generality, they
also become more stable, since attributes are defined by the persons they
represent. While the low utility, unstable attributes may change markedly
in meaning with each new stimulus person to which they are applied, high
utility attributes have Stable meanings Since they are defined by a vari-
ety of stimulus persons. Thus, the perceiver's judgments of stimulus per-
sons are more reliable and articulated when based upon higher utility
attributes.

The function of the developmental increase in attribute utility
is to increase the articulation of attribute judgments. In this sense,

the developing individual is becoming more competent in his perceptions

"8.31... ‘1; 3.. \.

 

49
of others; fine discriminations in the judgment of pe0ple are the prod-

uct of many previous experiences with people.

APPENDICE S

APPENDIX A

Free Response Description Coding System

Appearance

 

1. weight: "100 lbs."

A. "fat" "chubby"

B. "Skinny”

M. "bigger than me around the waist"
2. height: "five feet tall"

A. "tall"

B. "short"

M. "taller than me"
3. age: "42 years old”

A. "real old"

B. "young"

M. "two months older than me"
4. attractiveness

A. "pretty" "cute"

B. "ugly"

Roles and activities

 

5. occupation: "teacher" ”drives the bus works downtown”

M. "a teacher of mine"

6. athletic activities: "likes to play sports plays baseball" "goes

snowmobiling all the time"

M. "plays frisbee with my friends and me we play football"

50

. -5: 1.1;... Jilin!“
. l H“

7.

51
family position: "uncle of my friend" "Jane's dad"

M. "my mother” "my little sister"

Intellect and competence

 

8.

10.

ll.

12.

13.

intelligence

A. "intellectual" "smart" "knows a lot of stuff"

B. "dumb" "talks Stupid all the time”

M. "smarter than me"

school grades: "gets average grades"

A. "does well in school"

B. "he's always getting bad grades on his card”

A. good at his job: "great teacher" "ideal boss"

B. poor at his job: "worst principal we've ever had" "lousy
mechanic"

A. good at something: ”good football player" "talented singer"

B. poor at something: "can't fix cars"

A. ambitious: "motivated" "hard-working" "tries hard to keep up"
"busy"

B. lazy" "never gets his work done" "never does anything" "lazy
as hell"

A. immature: "babyish" "childish" "acts like a baby"

B. mature: ”more mature than most people her age"

Interpersonal traits

 

14.

15.

A. friendly: "congenial" "gets along with others" "nice to others"

"understands others" "makes people feel comfortable"
B. not friendly: "doesn't get along with others"

M. my friend: "friendly to me" ”my best friend" "my boyfriend
right now”

A. considerate: "kind” "respects others"

B. not considerate

16.

17.

l8.

19.

20.

21.

22.

23.

24.

52

M. considerate to me: "thinks of me first" "reSpects my feelings"

"kind to me all the time"
A. trustworthy: "has never broken a confidence”
B. not trustworthy
M. I trust him: "I would trust her anytime, anyplace"
A. generaous: "she isn't stingy like her parents"
B. selfish: ”he won't share things like he should"

M. generous to me: "always buys me things" "gives me things”
"won't ever let me play with his stuff”

A. helpful: "helps people with their problems" "he helps kids
when they don't know what to do"

B. not helpful

M. helps me: "he helped me through" "we talk about our problems
and she gives me good advice"

A. humorous: "funny" "witty"

B. not humorous

M. makes me laugh: "he kids around and makes me giggle"

A. good sense of humor: "laughs all the time"

A. conceited: "bragging" "self-centered" "thinks she's somebody
she's not" ”snobbish" "big head" "expects peOple to kiss his
feet"

B. not conceited
M. “thinks he's bigger than me"

H H

A. stubborn: "narrow-minded” ”set in his ways sticks to her

beliefs"
B. not stubborn
A. critical: "finds fault" "says when people do things wrong"
B. not critical
M. critical of me: "tells me I'm not doing things right"
A. gossips: "talks about her neighbors”

B. doesn't gossip

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

53
gossips about me

liar: "lies all the time" "tells stories I don't believe"
"exaggerates"

doesn't lie

lies about me: "tells my boyfriend things about me that just
aren't true"

outspoken: "loud" "loudmouth" "talks too much" "has a very big
mouth”

"know-it-all" "always thinks she knows best"

phony: "two-faced” "backstabbing" "different person to differ-
ent people"

not phony: "genuine” "Sincere"

bothersome: "pest" "bugs pe0ple all the time"

pesters me

show-off: "always showing off in front of the girls" "he's
good at sports and makes sure everyone knows when we play"

aggressive (verbally): "yells at people for nothing" "crabby"
"grouchy” "acts mean to everyone"
mean to me: "yells at me" "tries to scare me and telling me

what he's going to do"

aggressive (physically): "slugs people" ”hits girls" "breaks
things that aren't his" "bully”

hits me, breaks my stuff
always gets in trouble
always gets me in trouble

dominating: "wants things her own way" "never asks you, always
tells you" "can't wait to get his hands on power”

dominates me: "thinks he can tell me what to do"

hot temper: "gets mad over little things" "quick temper"
"always angry"

gets mad at me: "when I do stuff she doesn't like, she really
gets mad"
easy going: "you can talk together without him getting angry”

”never in a bad mood" "happy all the time"

37.

38.

39.

40.

54

cries: "cries always when she wants things" "pouts when he
can't have his way"

polite: "well-mannered" "a real gentleman"

polite to me

manipulating: "uses people" "makes people let her have her way"
likes kids: "understands kids" "relates to us like most adults
can't"

Relationship with Subject

 

41.

42.

43.

44.

45.

46.

M.

time Span of relationship: "known her since junior high" "I have
seen him every summer for the last few years" "we just met"

my neighbor: ”lives down the Street” ”lives next door"
we do things together: "we're always going places together"

we talk together

likes me: "enjoys my company”
Similar to me: "sees things the same as I do" "we are a lot
alike”

Global evaluations

 

47.

48.

49.

50.

51.

A.

nice person: "good person" "sweet guy" "fantastic"

not a nice person: ”weird" "asshole" "strange" "rotten old man"
I like the person

I dislike the person

I love the person

I hate the person

fun to be with: "we have fun together she's great to just be

with"
people like the person

pe0ple don't like the person

"x. ., I

APPENDIX B

Preliminary Analyses for Experiment 1

TABLE B1

Means and Standard Deviations of Variables
Assessed in Experiment 1

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Mean S.D.
Grade 8.07 3.10
Verbal intelligence 105.93 13.78
Differentiation score 30.10 11.24
Number of high utility attributes 6.27 3.69
Utility ratio .208 .0937
TABLE B2
Correlations Among Variables Assessed in Experiment 1

1 2 3 4 5

1. Grade - .12 .33 .62 .54
2. Verbal IQ - .39 .30 .03a
3. Differentiation - .57b -.04b
4. High utility attributes — .73b
NOTE: Variables 3, 4, and 5 are interdependent, thus the significance

of correlations among them is indeterminate. The subscript (b)
identifies these correlations. The criterion value of E: in all
other cases, is .27 for p_< .01. Correlations with subscript
(a) are not Significant.

55

APPENDIX C

Analyses of Differentiation Scores

TABLE Cl

Summary of Sex by Grade Unweighted Means Analysis
of Variance of Differentiation Scores

 

 

 

 

 

 

 

Source df MS F
Sex 1 895.35 9.22 (p < .005)
Grade 3 622.06 6.41 (p < .001)
Sex 5 Grade 3 139.99 1.44 ns
Error 80 97.09
TABLE C2

Summary of Sex by Grade Unweighted Means Analysis of
Covariance with Differentiation Scores as the
Criterion Measure and Verbal Intelligence
Scores as the Covariate

 

 

 

Source df MS F
Sex 1 721.66 8.49 (p < .005)
Grade 3 512.96 6.04 (p < .005)
Sex M Grade 3 107.87 1.27 ns
Error 79 84.99

 

 

 

 

56

... . a. I. )lvllibb. ....‘>Nd

a.

APPENDIX D

Analysis of Utility Ratios

TABLE D1

Summary of Sex by Grade Unweighted Means Analysis
of Variance of Utility Ratios

 

 

 

Source df MS F
Sex 1 .0052547 < 1
Grade 3 .0805704 13.23 (p < .001)
Sex 5 Grade 3 .0007039 < 1
Error 80 .0060881

 

 

 

 

57

“4; I

REFERENCES

REFERENCES

Attneave, F. Applications of information theory to psychology. New
York: Holt, 1959.

Attneave, F. Perception and related areas. In S. Koch (Ed.), Psychol-
ogy: A study of a science. Study 11, Vol. 4. New York:
McGraw-Hill, 1962.

 

Beach, L., and Wertheimer, M. A free reSponse approach to the study of
person cognition. Journal of Abnormal and Social Psychology,
1961, 92, 367-374.

Bieri, J. Cognitive complexity and personality deve10pment. In 0. J.
Harvey (Eds.), Experience, structure, and adaptability. New
York: Springer, 1966.

Bieri, J., Atkins, A. L., Briar, S., Leaman, R. L., Miller, H., and
Tripodi, T. Clinical and social judgment. New York: Wiley,
1966.

Biskin, D., and Crano, W. D. Structural organization of impressions
derived from inconsistent information: A developmental study.
E. Lansing:.,Michigan State University, 1973 (mimeo).

Bruner, J. S. On perceptual readiness. Psychological Review, 1957, pg,
123-152.

Bruner, J. S., Olver, R. R., and Greenfield, P. M. Studies in cognitive
growth. New York: Wiley, 1966.

Bruner, J. S., and Tagiuri, R. The perception of peeple. In G. Lindzey
(Ed.), Handbook of social psychology. Reading, Mass.: Addison-
Wesley, 1954.

Coombs, C. H., Dawes, R. M., Tversky, A. Mathematical peychology.
Englewood Cliffs, N.J.: Prentice-Hall, 1970.

Crockett, W. H. Cognitive complexity and impression formation. In
B. A. Maher (Ed.), Progress in experimental personality research,
Vol. 2. New York: Academic Press, 1965.

Crockett, W. H. Structural analysis of the organization of written
impressions. Unpublished manuscript, University of Kansas, 1970.

Dornbush, S., Hastorf, A. H., Richardson, S. A., Muzzy, R. E., and Vree-
land, R. S. The perceiver and the perceived: Their relative

58

59

influence on the categories of interpersonal perception. Journal
of Personality and Social Psychology, 1965, 1, 434-440.

 

Faguy-Coté, E. Cognitive centrality: Conceptualization and measurement.
Boulder: University of Colorado, 1965 (mimeo).

Flavell, J. H. The developmental psychology of Jean Piaget. Princeton,
N.J.: Van Nostrand, 1963.

Garner, W. R. Uncertainty and structure as psychological concepts. New
York: Wiley, 1962.

Gibson, E. J. Perceptual learning. Annual Review of Psychology, 1963,
14, 29-56.

 

Gollin, E. S. Organizational characteristics of social judgment: A
developmental investigation. Journal<foersonality, 1958, 26,
139-54.

Hastorf, A. H., Richardson, 8. A., and Dornbush, S. M. The problem of
relevance in the Study of person perception. In R. Tagiuri and
L. Petrullo (Eds.), Person perception and interpersonal behavior.
Stanford: Stanford University Press, 1958.

Kelling, G. Measuring cognitive integration. Unpublished manuscript,
University of Colorado, 1968.

Livesley, W. J., and Bromley, D. B. Person perception in childhood and
adolescence. New York: Wiley, 1973.

 

Lorge, 1., and Thorndike, R. L. Large-Thorndike Intelligence Tests.
Boston: Houghton-Mifflin, 1954.

 

Maccoby, E. E. The development of sex differences. Stanford: Stanford
University Press, 1966.

 

Neisser, U. Cognitive psychology. New York: Appleton-Century-Crofts,
1966.

 

Olshan, K. M. The multidimensional structure of person perception in
children. Unpublished Ph.D. dissertation, Rutgers University,
1970.

Peevers, B., and Secord, P. F. Developmental changes in attribution of
descriptive concepts to persons. Journal of Personality and
Social Psychology, 1973, 21, 120-128.

 

 

Restle, F. Psychology ofpjudgment and choice: A theoretical essay. New
York: Wiley, 1961.

Rosch, E. H. On the internal structure of perceptual and semantic cate-
gories. In T. E. Moore (Ed.), Cognitive development and the
acquisition of language. New York: Academic Press, 1973.

 

60

Rosenberg, S., and Sedlak, A. Structural representations of implicit
personality theory. In L. Berkowitz (Ed.), Advances in experimen-
tal social psychology. Vol. 6. New York: Academic Press, 1972.

 

 

Scott, W. A. Structure of natural cognitions. Journal of Personality
and Social Psychology, 1969, 11, 261-278.

 

 

Scott, W. A. Varieties of cognitive integration. Journal of Personality
and Social Psychology, in press.

 

 

Signell, K. A. Cognitive complexity in person perception and nation per-
ception: A developmental approach. Journal of Personality, 1966,
34, 517-537.

 

Terman, L. M., and Tyler, L. E. Psychological sex differences. In
L. Carmichael (Ed.), Manual of child psyehology. New York:
Wiley, 1954.

 

Ulehla, Z. J. Individual differences in information yields of raters.
Unpublished M.A. thesis, University of Colorado, 1961.

Vygotsky, L. S. Thought and language. Cambridge, Mass.: M.I.T. Press,
1962.

 

Wallach, M.' On psychological Similarity. Psychological Review, 1958, §§,
103-116.

 

Warr, P. B., and Knapper, C. The perception of people and events, New
York: Wiley, 1968.

 

Winer, B. J. Statistical principles in experimental design. New York:
McGraw—Hill, 1971.

 

Wohlwill, J. F. From perception to inference: A dimension of cognitive
development. In W. Kessen and C. Kuhlman (Eds.), Thought in
the young child. Monogrephs of the Society for Research in Child
Development, 1962, 21, 87-112.

 

 

Yarrow, M. R., and Campbell, J. D. Person perception in children.
Merrill-Palmer Quarterly, 1963, 2, 57-72.

Zajonc, R. B. Cognitive theories in social psychology. In G. Lindzey
and E. Aronson (Eds.), Handbook of social psychology, Vol. 1,
Reading, Mass.: Addison-Wesley, 1968.

 

E

”"llfillljllljmu

3

     

filjljljilillllll“