pected to better master new and unsolved circumstances, to maintain integrated .... airplane at the top, a teddy bear in the middle, and a clown at the bot- tom.
Journal of Personality and Social Psychology 1986, Vol. 51, No. 2,423-434
Copyrighl 1986 by the American Psychological Association, Inc. 0022-35l4/86/$00.75
Egocentrism and Ego Resiliency: Personality Characteristics Associated With Perspective-Taking From Early Childhood to Adolescence Per F. Gjerde, Jack Block, and Jeanne H. Block University of California, Berkeley In this study we examined the longitudinal relation between ego resiliency and egocentrism. Measures of Level 1 and Level 2 egocentrism were administered to one hundred eleven 3'/2-year-old children. Independent personality evaluations were available on these children at ages 3, 4, 7, 11, and 14 in the form of Q-sort ratings by independent sets of teachers and examiners. The relation between egocentrism and ego resiliency was observed to depend on level of egocentrism, as well as on sex of subject. In the sample of girls, both Level 1 and Level 2 egocentrism were negatively correlated with ego resiliency concurrently at age 3, but no long-term implications of egocentrism emerged. In the sample of boys, Level 1 egocentrism was consistently negatively associated with ego resiliency from early childhood (age 3) into adolescence (age 14). However, Level 2 egocentrism displayed no concurrent nor any dependable longitudinal relation with ego resiliency. The discussion focuses on possible interpretations of the relation between egocentrism and ego resiliency and on the sex differences in the pattern of longitudinal results.
As Piaget (1928, 1950) has noted, the very young child has
children's perspective-taking abilities (Ravell, 1978). Indeed,
difficulties seeing viewpoints other than its own. In cognitive-
rudimentary perspective-taking skills apparently emerge as
developmental theory, this characteristic of egocentrism is
early as at the age of 18 months (Masangkay et al., 1974).
viewed as dominant in young children's thought (Piaget & In-
Conclusions about the normative decline in egocentrism
helder, 1956). Denned as the inability to differentiate one's own
based on cross-sectionally obtained group means may be mis-
viewpoint from that of others, egocentrism is manifested by a
leading. Even at a very young age, children's perspective-taking
failure to perceive an object in more than one way when alterna-
skills display striking individual differences. Studying 4-year-
tive perspectives are both possible and necessary. Egocentrism
old children, Light (1979) reported a considerable range in per-
depends upon and is controlled by what is directly perceived,
spective-taking scores. With a possible maximum of 40, scores
requiring no conceptual elaborations. Perspective-taking, or
ranged from 9 to 37, with a mean of 20; this indicates that some
nonegocentrism, requires detachment from and cognitive ma-
4-year-olds already had achieved an adequate understanding of
nipulation of perceptual experience.
different perspectives, whereas others had remained essentially
Studies of egocentrism reflect Piaget's theoretical emphasis
egocentric. (See also Flavell et al., 1968, for additional evidence
on the common forms of structural development and a develop-
of individual differences in perspective taking.)
mental-descriptive research orientation (e.g., Ravell, Botkin,
The antecedents of individual differences in egocentrism are
Fry, Wright, & Jarvis, 1968). Such studies have focused primar-
not well understood, partly because the origins of age-related
ily on the age-related decline of egocentrism, on the relation
decline in egocentric thought have received little or no atten-
between egocentrism and other areas of cognitive functioning,
tion. From a normative perspective, the decline of egocentrism
and on the interrelation between different domains of egocen-
with age has traditionally been explained by reference to social
trism. (See Ford, 1979, and Shantz, 1983, for reviews of this
interaction. In his early writings, Piaget argued that the child
research.) According to Piaget's (1928) early research on this
achieved nonegocentric thought through interpersonal experi-
topic, children do not achieve nonegocentric thought until the
ences (e.g., Piaget, 1928, 1950; see also Flavell, 1963). He sug-
late-preschool/early-school-age period. However, more recent
gested that negative reinforcement from peers forced the child
findings indicate that Piaget may have underestimated young
to reexamine his percepts in light of those of others. Later, however, Piaget attributed less unequivocally a causal role to social experiences (e.g., Piaget & Inhelder, 1969). According to his
This study was supported by National Institute of Mental Health Grant MH 16080 awarded to Jack and Jeanne H. Block. Per F. Gjerde was supported by Grant B.68.80.006 from the Norwegian Research Council for Science and the Humanities. This article benefited from comments by Kjell Doksum, Rachel Melkman, Margaret Ricks, and Howard Terry. Correspondence concerning this article should be addressed to Per F. Gjerde or to Jack Block, Department of Psychology, Tolman Hall, University of California, Berkeley, California 94720.
more recent argument, thinking and social behavior develop together in a manner that makes it difficult to separate cause from effect. Thus, the question of the direction of influence is left more open (cf. Light, 1979). Empirical research does not unequivocally support the causal role of interpersonal experience. Comparing rural and urban children, Hollos and Cowan (1973; Hollos, 1974) found that isolated children performed less well than village and town
423
424
P. GJERDE, J. BLOCK, AND J. BLOCK
children on perspective-taking tasks. However, perspective-tak-
(brittle) individual lacks this ability to respond dynamically to
ing abilities were not a regular function of the amount of social
changing environmental circumstances. In addition to serving
interaction. For example, although children living in towns had
specific cognitive ends, ego resiliency also serves broader moti-
more experience in social interaction than children living in
vational and affective functions.
villages, their perspective-taking scores were identical. Hollos
As an instance of a more general ability to adapt resource-
and Cowan advanced a threshold model in which a minimum
fully to changing environmental contingencies, perspective-tak-
of social experience is a necessary prerequisite for developing
ing competence in early childhood can be expected to be posi-
perspective-taking competence to explain these findings. Once
tively related to concurrently assessed ego resiliency. In addi-
this minimum level has been reached, however, additional op-
tion, early attainment of perspective taking should foretell later
portunity for social interaction has little effect. Nahir and Yus-
adaptation. In particular, we expect children who can see view-
sen (1977) found differences in role-taking abilities favoring Is-
points other than their own to be more likely to adapt resource-
raeli kibbutz children over city children. Light (1979), on the
fully (i.e., to be resilient) in a variety of contexts through middle
other hand, reported that the children who spent the most time
childhood and into adolescence. Preliminary evidence for this
with peers did not score higher than other children on perspec-
relation has been presented by Light (1979), who found that
tive-taking tasks. Similarly, West (1974) found no differences in
competent roletakers accommodated more easily to new rela-
role-taking abilities among Israeli children living in three
tionships and situations, such as the school experience.
different communities (city, moshav, and kibbutz). Conse-
The study reported here used a longitudinal design to exam-
quently, at this time it does not seem possible to formulate clear
ine the relation between egocentrism and ego resiliency in a
conclusions about the direction of influence between perspec-
sample of children who were assessed independently over a pe-
tive taking and opportunity for social interaction.
riod of 11 years, from early childhood (age 3) into adolescence
We suggest that inquiries into the antecedents of individual
(age 14). A longitudinal design is especially appropriate because
differences in egocentrism should pay more attention to chil-
prior research on egocentrism has neglected to assess its devel-
dren's personality-motivational characteristics. Cognition and
opmental trajectory on an individual level. We are aware of only
personality are known to be connected; yet most contemporary
one other study that has examined the longitudinal implications of egocentrism. Light (1979) reported that children who per-
examinations are compartmentalized and neglect to study how personality characteristics may influence cognitive functioning.
formed well on role-taking tasks at age 4 were socially more
Although a priori, the construct of egocentrism appears to exist
confident and independent at age 3, and were less likely to show
at the intersection of personality and cognition, the concept has
maladjustment later in elementary school. The study reported
not been examined in the context of theories that stress the mo-
here was conducted to extend the investigation of the develop-
tivational basis for cognitive development (e.g., J. Block, 1982;
mental implications of early preschool egocentrism into the ad-
Rapaport, 1951). Nonegocentric inferences, for example, re-
olescent years.
quire more than the mere cognitive ability to anticipate what
Our view on egocentrism derives from an important series of
other persons can or cannot see. An additional and necessary
articles by Flavell (1974,1978; Flavell, Everett, Croft, & Flavell,
component is the perceived need to accommodate other per-
1981;
spectives in social interaction (Flavell, 1974). As Light (1980)
Shipstead, & Croft, 1978; Lempers, Flavell, & Flavell, 1977;
Flavell,
Flavell, Green,
& Wilcox,
1980; Flavell,
has remarked, a motivationally based neglect of other view-
Masangkay et al., 1974), which describe children as undergoing
points is frequently implied when someone is referred to as ego-
systematic changes in their abilities to predict another's visual
centric. This attribute suggests a gap between inferential abili-
perspective. According to Flavell, visual perception develops in
ties and their actual application. Reflecting on his findings,
sequenced levels of knowledge and progresses from a less ad-
Light (1979) concluded that the striking variability observed in
vanced state (Level 1) to a more advanced state (Level 2). At
perspective-taking scores probably stems less from differences
Level 1, the child may be able to infer the particular objects
among children in absolute cognitive-developmental level than
another person does or does not see. However, although the
from variations in children's general awareness of differences in
child may infer accurately the particular objects another person
perspectives and the need to adapt to them.
may be seeing, he or she is not yet aware of differences in the
The current study examined individual differences in early
visual perspectives of various observers. That is, the child does
childhood egocentrism from a personality viewpoint. We con-
not yet understand that an object simultaneously visible to both
sidered the attainment of nonegocentric thought as an age-rele-
self and others may nevertheless appear differently when viewed
vant instance of a more general ability to adapt resourcefully to
from another position. At Level 2, on the other hand, the child
changing environmental contingencies—a personality characteristic we have labeled ego resiliency (J. Block, 1965, 1982,
has acquired the crucial additional notion of a position-deter-
J. H. Block & J. Block, 1980). In our conceptualization, ego
tions in visual experiences are possible even when the same ob-
resiliency refers to an individual's dynamic capacity to modify
jects are simultaneously visible to different observers. This
his or her modal behavior in the face of changing environmental
Level 2 competence is precisely what Piaget and Inhelder (1956)
demand characteristics. The ego-resilient person can be ex-
and most subsequent researchers have examined in studies of
pected to better master new and unsolved circumstances, to
perspective taking, and this ability is what the classical three-
maintain integrated performance under stress, to process com-
mountain procedure was designed to assess. In the present
mined perspective. Children at this level now realize that varia-
peting stimuli, to resist sets and illusions, and to be less immobi-
study, early preschool children were assessed with respect to
lized, repetitive, and anxious under stress. The ego-unresilient
both Level 1 and Level 2 competence.
EGOCENTRISM AND EGO RESILIENCY
Method Subjects The initial subject sample included 111 children, 53 boys and 58 girls, participating in an ongoing longitudinal study of ego and cognitive development initiated in 1968 at the University of California, Berkeley, by Jack and Jeanne Block. (For a comprehensive description of this study, see J.H. Block &J. Block, 1980.)The exact number of subjects in any given analysis varied somewhat. About two thirds of the subjects were white, one quarter were black, and one twelfth were Asian. Subjects were initially recruited into the study at age 3, while attending either a university-run nursery school or a parent cooperative nursery school, and were assessed on wide-ranging batteries of measures at ages 3,4, 5 , 7 , 1 1 , and 14. The children, now adolescents, are currently being assessed at age 18. Subjects live primarily in urban settings and are heterogeneous with respect to social class and parent education.
Egocentrism Tasks When the children were 3 years old, a wide-ranging battery of tests, including measures of Level 1 and Level 2 egocentrism, was administered during different evaluation sessions. Level 1 egocentrism procedure. The Level 1 task is similar to the test designated as HIE by Flavell et al. (1968). It was designed to assess the child's ability to predict the objects that other persons can or cannot see, without requiring the child to demonstrate knowledge of positiondetermined perspectives. On each side of an 11-in.X 14-in.(28-cmx35.5-cm) board, the same three colored pictures in the same spatial positions were displayed: an airplane at the top, a teddy bear in the middle, and a clown at the bottom. In addition, two pieces of plain, red cardboard, each 3 in. (7.5 cm) wider than the board, were provided. One of these pieces was hinged in the middle. The child and the experimenter were seated facing each other across a small table. The experimenter made sure that the child could see that the same pictures in the same spatial order were back-toback. During the pretrial position, the experimenter used the red, folded cardboard to cover two of the pictures on both sides, held the board upright, and asked the child to guess what the experimenter could see on her side of the board. The actual test consisted of four trials. On each, the experimenter covered a different part of the board. On the first, she covered one picture (airplane) on her own side. On the second trial, she covered two pictures (airplane and bear) on her side. On the third trial, the experimenter covered one picture (airplane) on child's side and two pictures (airplane and bear) on her side. On the final trial, the experimenter covered all three pictures on her side. Each time, the child was asked what the experimenter could see. An egocentric response was defined as one in which the child chose a picture that could be seen from his or her side but not from the examiner's side, because it was covered with cardboard. The child was given a score of 1 for an egocentric answer and a score of 0 for a nonegocentric answer. The scores for the four trials were summed to provide the total Level 1 egocentrism score (possible range from 0 to 4). Errors unrelated to the dimension of egocentrism (e.g., selection of pictures that could be seen neither from the child's nor from the examiner's side) were not included in the egocentrism score. Level 2 egocentrism procedure. The Level 2 task—a variation of the three-mountains procedure—was designed to assess the child's knowledge or recognition of the possibility of a position-determined perspective, the defining characteristic of Level 2 perspective-taking competence. The child was presented with a toy elephant, a wooden bird, a toy clown, a set of three pictures showing front, side, and back views of the elephant, and another set of three pictures showing front, side, and back
425
views of the bird and the elephant together. The task consisted of six trials preceded by a practice trial. On each trial, the relative position of the toys differed. On the first three trials, only the elephant and the clown were included. On the last three trials, the bird was added. The child was asked to select the picture that showed the clown's view. Because this task presented the child with three options to choose among—his or her own view, the clown's view, and a third view—two types of errors were possible: (a) egocentric errors (the child chose his or her own perspective) and (b) errors unrelated to the dimension of perspective taking (the child chose neither his or her own perspective, nor that of the clown). Only egocentric errors were included in the total egocentrism score. Egocentric answers were given a score of 2, the scores on the six items being summed to provide the total Level 2 egocentrism score (possible range from 0 to 12).
Personality and Intelligence Measures Measuring intelligence. The Wechsler Preschool and Preliminary Scale of Intelligence (WPPSI; Wechsler, 1967) was administered when the children were 4 years old. For girls, the mean scores were 117.791 (SD = 12.91) for Verbal IQ; 116.24 (SD = 12.70) for Performance IQ; and 118.94 (SZ> = 12.02) for Full Scale IQ. For boys, these scores were 114.87 (SD= 14.32), 112.81 (SD = 12.37), and 115.43 (SD = 12.51), respectively. These differences in mean intelligence scores do not begin to approach statistical significance. Measuring personality: The California Child Q-set. The children's personality characteristics were described by their nursery school teachers at age 3 and at age 4, by their public school teachers and project examiners at age 7 and at age 11, and by project examiners at age 14, using the standard vocabulary of the California Child Q-set (CCQ; J. Block & J.H. Block, 1969,1980; J.H. Block &J. Block, 1980;J.Block, J.H. Block, & Harrington, 1974). The CCQ is an age-appropriate modification of the California Q-set (J. Block, 1961, 1978) and consists of 100 (or 63—see ahead) widely ranging statements about the children's personality, cognitive, and social characteristics. At age 3, all of the children were described by three nursery school teachers who had worked with them for a minimum of 5 months before completing the descriptions. The teachers also received training in the Q-sort methodology and met with the project director, who explained the rationale, provided written instructions to the CCQ, and answered questions about item meanings. The teachers then independently did a Q-sort for a child who was not in the study (usually one from a previous year) but who was known to all of the teachers. The item descriptions were discussed, and usually a second child was described to check the understandings. At age 4, each child was again described via the CCQ procedure, but by an entirely different set of three nursery school teachers. These teachers were equivalently trained. When the children were at age 7 and in public school, one teacher and two examiners provided the Q-sort characterizations. When the children were 11 years old, each child was described by four or five examiners who had observed him or her while administering a variety of experimental procedures tapping different aspects of cognitive and personality functioning. Similarly, when the children were 14 years old, each child was described by four examiners who had observed the child during experimental sessions. When the children were brought in for assessment at ages 7, 11, and 14, the examiners evaluating the children Q-sorted them, using only 63 of the original 100 CCQ items. The 37 CCQ items excluded were those for which the examiners believed reliable judgments could not be formulated, given the necessary constraints set by the laboratory environment. To ensure commensurativeness of the Q-data, as in earlier assessments, the examiners used a nine-step rectangular distribution in sorting the 63 CCQ items. In addition, examiners also described each
426
P. GJERDE, J. BLOCK, AND J. BLOCK
Table 1 CCQ Items Most Positively and Most Negatively Defining Ego Resiliency Items most characteristic of ego resiliency
Items most uncharacteristic of ego resiliency
Vital, energetic, and lively (9.0) Resourceful in initiating activities (9.0) Curious and exploring (9.0) Self-reliant and confident (9.0) Perceptually creative (8.7) Recovers after stressful experiences (8.7) Responds to humor (8.3) Attentive and able to concentrate (8.3) Competent (8.3) Uses and responds to reason (8.3) Has an active fantasy life (8.0)
Rigidly repetitive under stress (1.0) Inappropriate in emotive behavior (1.0) Goes to pieces under stress (1.0) Inhibited and constricted (1.3) Feels unworthy, thinks of self as "bad" (1.3) Anxious in unpredictable environment (1.7) Fearful and anxious (1.7) Sulky and whiny (1.7) Bodily symptoms as function of tension (1.7) Withdraws under stress (1.7) Overreacts to minor frustrations (1.7)
Note. CCQ = California Child Q-set. The values in parentheses reflect mean placement of CCQ item in the ego-resiliency composite. Possible range = most uncharacteristic (1) to most characteristic (9).
individual at age 14, using the 100 items of the original, adult form of the California Q-set (J. Block, 1961,1978). Judges described each child by arranging the Q-set items in a forced nine-step, rectangular distribution, according to the evaluated salience of each item with respect to a particular child. They worked independently of each other. At each age, the independent Q-sort formulations were averaged to form composite Q-sort descriptions. The CCQ descriptions were completed by a total of 11 different nursery school teachers when the children were at age 3; an entirely different set of 9 nursery school teachers completed the sorts when the children were at age 4; 67 different public school teachers and 2 examiners offered their personality evaluations when the children were at age 7; 5 different examiners offered Q-sort formulations when the children were 11 years old; and still another set of 4 examiners provided Q-sort descriptions when the children were 14 years old. Thus, the assessments at each time period are strictly independent of each other and of the egocentrism measures. The estimated internal consistency reliabilities of the Qitems, based on correlations among observers, averaged .65 at ages 3 and 4. At age 7, the average item reliability was .47; at 11, it was .70; and at 14, it was .72. The relatively low average item reliability at age 7 is partly due to having used only 3 raters at this age level. This reliability information not only provides important information about the relative quality of the personality data used, but it also places perspective on the possible magnitude of correlations that can be expected when these measures are related to our criterion variable, egocentrism. It is insufficiently recognized that obtained correlations may be considerably attenuated as a function of the inevitable unreliability of the measures used (J. Block, 1963,1964; Epstein, 1979,1980). Operationally defining the ego-resiliency construct. Indexing the construct of ego resiliency was accomplished in two steps. The personality characteristics considered to be associated with the construct of ego resiliency were specified, beforehand, by three clinical psychologists who used the CCQ to describe, independently, a prototypical ego-resilient child. The criterion defmers showed high levels of agreement, the reliability of the criterion definition being .90. To convey a sense of the personality characteristics especially denning ego resiliency, Table 1 lists the 11 items that most positively define the concept, as well as the 11 items that most negatively define the concept. The ego-resiliency prototype score used in this study, however, is based on all 100 CCQ items. The second step toward creating an ego-resiliency score for each subject involved correlating the composited CCQ description of each child in the study with the ego-resiliency prototype. For each child, the correlation or congruence between his or her Q description by teachers or examiners and the ego-resiliency prototype was taken as a score, index-
ing the similarity between the child's personality and the construct of ego resiliency. A high correlation meant that the child was similar or close to the prototypical definition (i.e., resilient); a low or negative correlation meant that the child was dissimilar or far from the prototypical definition (i.e., brittle).
Results Reliabilities of the Level 1 and the Level 2 Egocentrism Tasks The measurement reliability of egocentrism procedures has been reported only infrequently; hence, our analyses may be of general interest. In our sample, the internal consistency (coefficient alpha) reliability of the Level 1 egocentrism measure was .79 for girls, and .70 for boys. For the Level 2 measure, the alpha reliability coefficients were .70 and .45 for girls and boys, respectively. Except for this last reliability coefficient, the reliabilities are relatively high for such short measures. In the subsequent analyses to be reported, the reader should recognize that the relations observed have been attenuated by the unreliability present in these (and other) measures (J. Block, 1963, 1964). In particular, the low reliability of the Level 2 measure in the male sample limits efforts to connect this measure with others.
Descriptive Statistics for the Egocentrism Measures Most 3-year-old children performed adequately on the Level 1 procedure. Forty-one of 58 (70%) girls and 34 of 53 (64%) boys achieved a perfect Level 1 score. Conversely, only 13 of 57 (23%) girls and 10 of 50 (20%) boys performed without error on the Level 2 procedure. This difference in Level 1 and Level 2 performance reaffirms (as is developmental^ predictable) that 3-year-olds have considerably greater difficulties making Level 2 judgments than Level 1 judgments. The number of children receiving each score on the two procedures is shown separately for each sex in Table 2. The mean Level I egocentrism scores were .60 (SD = 1.11, N = 58) for girls, and .68 (SD = 1.12, TV = 53) for boys. For Level 2 egocentrism, the mean scores were 4.11 (SD = 3.45,
EGOCENTRISM AND EGO RESILIENCY
427
Table 2
tion coefficient is inappropriate in the present context because
Score Distributions on Level 1 and Level 2 Egocentrism Tasks
the different egocentrism procedures represent hierarchically organized levels of knowledge. There are four possible combi-
No. of items incorrect Task
0
1
2
3
4
nations of Level 1 and Level 2 performance: Level 1 high ego5
6
centrism-Level 2 high egocentrism, Level 1 high egocentrismLevel 2 low egocentrism, Level 1 low egocentrism-Level 2 high
Girls Level 1 Level 2
41 13
7 12
4 11
egocentrism, and Level 1 low egocentrism-Level 2 low egocen-
4 9
2 7
_
_
trism. Of these, one combination—Level 1 high egocentrism-
2
3
Level 2 low egocentrism—is incompatible with a hierarchical model of perspective-taking development. Subjects falling in
Boys Level 1 Level 2
34 10
9 16
6 9
any of the three remaining combinations are manifesting re-
1 5
3 7
_
_
1
2
№>(e. For girls, Ns are 58 for Level 1 scores and 57 for Level 2 scores. For boys, Afe are 53 for Level 1 scores and 50 for Level 2 scores.
sponse patterns compatible with this model. To provide a conceptually more appropriate test of the Level 1-Level 2 developmental hypothesis, we treated performance on the Level 1 and Level 2 tests within a Guttman scale framework. Guttman scale analysis simultaneously orders both the items being scaled and the individuals being evaluated. The items are ordered by degree of difficulty, and the respondents
N = 57) for girls, and 3.76 (SD = 3.25, N = 50) for boys. Due
are ordered so that those who pass a difficult item tend to pass
to the highly skewed score distributions, the t test is inappropri-
less difficult items. This rationale for measurement is conceptu-
ate for testing the significance of sex differences. Instead, the
ally appropriate developmentally; subjects who make correct
Mann-Whitney test was used. The results showed no signifi-
Level 2 inferences should always be able to make correct Level
cance differences between egocentrism-taking scores of girls
1 inferences, but not vice versa.
and boys, a finding consistent with previous research (e.g., Kurdek, 1977;Turnure, 1975). Inspection of Table 2 shows that the distribution of children's
When we used the score of 4 (i.e., perfect performance) on the four-item Level 1 task and the score of 12 (i.e., perfect performance) on the six-item Level 2 task as the criteria for ade-
scores on the two egocentrism measures is highly skewed, espe-
quate performance, the data proved to be highly scalable in the
cially on the Level 1 task. Such skewedness introduces several
Guttman sense, thus supporting the Level 1-Level 2 hypothe-
problems of analysis and interpretation. In particular, when
sis. (The coefficient of reproducibility was .93; the minimum
variables with radically different score distributions are corre-
marginal reproducibility was .65; and the coefficient of scalabil-
lated, the subsequently computed product-moment corre-
ity was .79.) No sex differences emerged in scalability.
lations no longer have the conventionally expectable limits of
Previous studies have reported mostly unimpressive intercor-
±1.00. Rather, the range of possible values is less than ±1.00,
relations on performance among measures of egocentrism, thus
sometimes appreciably less. This frequent pulling in of limits,
casting doubt on the validity of the underlying construct (cf.
coupled with a general failure to recognize that correlations
Ford, 1979). But because, as we have shown here, low task inter-
may exist on a compressed scale, means a proper perspective on
correlations need not preclude lawful and theoretically expect-
the relations observed often may be lacking. That is, an ob-
able relations, future analyses of the integrative usefulness of
tained correlation of .30 should take on a different implication
egocentrism should carefully consider the developmental level
if that value is found in a metrical context limiting the maxi-
of the measures used before drawing conclusions on the basis
mum possible correlation to a value of .50 rather than 1.00. (For
of task intercorrelations.
further understanding of this analytical and evaluative problem, the reader should consult Carroll, 1961.) Note that this problem is separate from the attenuation problem due to measurement unreliability noted before. In the following, the meaning of the reported Pearsonian coefficients should be evaluated in light of their maximum possible value.' Hence, for the most important analyses, this maximum possible coefficient is reported. As will be seen shortly,
Relations Between Egocentrism and Intelligence For girls, the correlations between Level 1 egocentrism and WPPSI-measured intelligence were -.24
(ns) for Verbal IQ;
-.25 (ns) for Performance IQ; and -.27 (p < . 10) for Full Scale IQ (Ns range from 40 to 41). For this sex, the maximum possible correlations between this measure of egocentrism and the three
this value is often appreciably less than ± 1.00, especially in analyses involving the Level 1 task.
-.15 (ns) for girls, and .20 (ns) for boys, a difference in corre-
' The maximum possible product-moment correlation coefficient is readily calculated (Carroll, 1961). One way involves first reordering the set of scores for the X variable in descending order; likewise, the scores for the Y variable are reordered in descending order. If these two reordered vectors of scores are then correlated in the usual way, the maximum possible correlation will result If one of these variables is reordered in ascending order and the other variable remains reordered in descending order, the subsequent correlation will provide the maximum
lations that is not significant. However, the logic of the correla-
possible negative correlation.
Developmental Implications of Level 1 and Level 2 Egocentrism Some readers may be interested in knowing that the correlation between Level 1 and Level 2 egocentrism performance is
428
P. GJERDE, J. BLOCK, AND J. BLOCK
measures of intelligence were, in order: -.77, -.73, and -.73. For boys, these correlations were -.27 (ns), -.03 (ns), and -.23 (ns), respectively (№ range from 35 to 36). For this sex, the maximum possible correlations were -.85, -.80, and -.84, respectively. Thus, children who manifest Level 1 egocentrism tend to perform somewhat less well on this test of intelligence; however, the magnitudes of these relations are small, even when considered in light of the compressed scale on which the coefficients fall. For girls, Level 2 egocentrism showed no relations to intelligence approaching significance, the correlations ranging from - .08 to -. 11. (Maximum possible correlations range from -.92 to -.95.) For boys, the correlations between Level 2 egocentrism and intelligence were .24 (ns) for Verbal IQ; .47 (p < .01) for Performance IQ; and .45 (p< .01) for Full Scale IQ. (Maximum possible correlations range from .95 to .97.) That is, Level 2 egocentrism is, surprisingly, positively related to IQ in boys. Egocentrism Related to the CCQ Ego-Resiliency Prototype The Q-sort derived indices of ego resiliency, based upon independently obtained CCQ evaluations at ages 3,4,7,11, and 14, were correlated separately with the Level 1 and the Level 2 tasks. The results for the sexes, considered separately and combined, are reported in Table 3. Due to the skewed score distributions for the two egocentrism measures, we have included for each task, and separately by sex, the maximum possible correlations in order to provide additional understanding of the observed relations. As seen in Table 3, the maximum possible correlations between egocentrism and ego resiliency differ somewhat across the five age levels. This recognition introduces a problem in comparing the egocentrism/ego-resiliency correlations across age levels. To create a common metric, each observed correlation coefficient was divided by its maximum possible value. This procedure is intuitively sound and, in addition, provides an estimate of what the correlation coefficient would have been had it fallen on a scale ranging from 1.00 to -1.00 (Cureton, 1959). Within the female sample, the Level 1 and Level 2 measures of egocentrism are negatively related to resiliency concurrently at age 3, but consistent subsequent longitudinal relations do not emerge. Within the male sample, the relation between Level 1 egocentrism and ego resiliency is concurrently and longitudinally consistently negative; all five correlations are in the expected direction, and three are statistically significant (p < .01, two-tailed). Level 2 egocentrism, however, is unrelated to ego resiliency in boys. For boys, the longitudinal relations between Level 1 egocentrism and ego resiliency are impressive directly and especially when the magnitude of the observed relations are viewed in light of their maximum possible size. These sex-differentiated relations between egocentrism and ego resiliency do not appear to derive from sex differences with regard to the standard deviations of the egocentrism and the ego-resiliency scores. Moreover, the intercorrelations of the egoresiliency scores from age 3 to age 14 are appreciable and not
different for the two sexes. Hence, the differences in the relations observed seem to derive primarily from differences between the sexes in the characterological implications of egocentrism. Although the correlations between Level 1 egocentrism and WPPSI-measured intelligence were low, we sought to evaluate the possibility that the ego-resiliency correlates of Level 1 egocentrism in the male sample were influenced by intelligence level. Partialing out WPPSI Full Scale IQ had only a minor effect on the longitudinal correlations between Level 1 egocentrism and ego resiliency. The longitudinal correlations at the five age levels were —.47 (p < .01), —.20 (ns), —.26 (ns), —.48 0>