A Critique of the Proposed National Education Policy Reform
Â
A Self-Image Questionnaire For Young Adolescents (SIQYA) Reliability And Validity Studies
1. Journalof Youthand Adolescence, t/ol. t3. No. 2, 1984
A Self-Image Questionnaire for Young
Adolescents (SIQYA): Reliability and
Validity Studies
Anne C. Petersen, ~ John E. Schulenberg, z
Robert H. Abramowitz, 3 Daniel Offer, 4 and Harold D. Jarcho 5
ReceivedFebruary1, 1984,"acceptedFebruary20, 1984
This article describes a questionnaire measure of setf-hnage designed for
young adolescents. It represents a downward extension of the Offer Self-
hnage Questionnaire and utilizes nine scalesfrom that instrument: Emotional
Tone, Impulse Control, Body Image, Peer Relationships, Family Relation-
ships, Mastery and Coping, Vocational/Educational Goals, Psychopathol-
ogy, and Superior Adjustment. This 98-item questionnaire elicits responses
on a 6-point Likert-type scale. The alpha coefficientsfor each scale are high,
indicating a high degree of internal consistency among the items. The validi-
ty of this instrument is examined through factor analyses and through the
association of these scales with other measures of self-hnage. The results sug-
gest that this questionnaire provides a useful way to assess self-image among
young adolescents.
This research was supported by Grant MH 30252/38142 to Anne Petersen.
~Professor of Human Development and Head, Department of Individual and Family Studies,
The Pennsylvania State University. Received Ph.D. from the University of Chicago in 1973.
Research interest is biopsychosocial development in adolescence, with a focus on sex differences.
:Doctoral candidate, Human Development and Family Studies, The Pennsylvania State Univer-
sity. Research interests are vocational development and contextual influences on development
in early adolescence.
3Doctoral candidate, Human Development and Family Studies, The Pennsylvania State Univer-
sity. Research interests are the influences of adolescent and parent development on family rela-
tionships.
~Professor of Psychiatry, University of Chicago, and Chairman, Department of Psychiatry,
Michael Reese Hospital and Medical Center. Received M.D. from the University of Chicago.
Research interests are the psychology and psychopathology of adolescents.
5Doctoral candidate, Educational Psychology, University of Chicago. Current research interests
are social cognition and peer relationships during early adolescence.
93
()1)47-2891f8410400-0093S03.50/0 ~ 1984Plcnunl PublishingCorporation
2. 94 Petersen, Schulenberg, Abramowitz, Offer, and Jarcho
INTRODUCTION
The purpose of this article is to provide information on the psycho-
metric properties of the Self-lmage Questionnaire for Young Adolescents
(SIQYA). The SIQYA, an adaptation of the Offer Self-Image Questionnaire
(OSIQ; Offer et al., 1981), is designed to measure several dimensions of self-
image in young adolescents. Early adolescence, characterized by rapid bio-
logical, psychological, and social changes, is a phase of development dis-
tinct from the later years of adolescence (e.g., Hamburg, 1974). Differences
in the cognitive and psychological functioning of young adolescents are par-
ticularly important to the assessment of self-image at this age. Hence, as we
found in our research on 11- 13-year-olds, questionnaires designed to meas-
ure self-image in older adolescents may not be entirely appropriate for youn-
ger adolescents. In this article we describe the development of the SIQYA
and present reliability and validity results obtained in two cohorts of young
adolescents. We begin with a brief discussion of the concept of self-image
and self-image development in adolescence to provide a context for the is-
sues relevant to the measurement of self-image among young adolescents.
SELF-IMAGE
Self-image, or self-concept, 6 is essentially a phenomenological organi-
zation of individuals' experiences and ideas about themselves in all aspects
of their lives (Coombs, 1981). Self-image is manifested through functioning
in various social domains (e.g., school, family, peer group), as well as through
psychological functioning (e.g., impulse control, m~ntal health adjustment,
ease in new situations) (Offer et al., 1981). Therefore, we assume that self-
image is multidimensional and should be measured as such (e.g., Dusek and
Flaherty, 1981; Offer and Howard, 1972; Wylie, 1974). Self-esteem, defined
as a generalized sense of self-worth or self-acceptance (Wylie, 1979), is simi-
lar to but not the same as self-image. Rather, self-esteem is viewed as the
evaluative component of self-image (Rosenberg, 1979; Wells and Marwell,
1976) and is usually considered unidimensionally (see Abramowitz et al., in
preparation).
According to Mead (1934), and articulated particularly with respect to
adolescents by Rosenberg (1979), self-image develops primarily through so-
cial interactions in which the self-image changes in response to discrepancies
between one's perceived self-image and others' reactions. While this view iden-
6in the literature,distinctionsare not drawn betweenthe terms"self-image"and "self-concept,"
and both terms appear to be used to identify the sameconstruct. Hence, the two terms are
used interchangeablyin this paper.
3. MeasuringSelf-Image of YoungAdolescents 95
tifies contextual factors as providing the impetus for self-image development,
a complementary perspective focusing on ontogenetic facts is equally impor-
tant. Developmental change is especially essential to consider during early
adolescence, given the biological and cognitive changes occurring at this time
(e.g., Petersen and Spiga, 1982). For example, one would expect self-views
related to physical appearance to change as a result of the onset of puberty
during early adolescence. Hence, our model for considering the development
of self-image is a transactional one (Sameroff, 1977), in which development
occurs as a result of interactions between ontogenetic and contextual factors
(Petersen, 1981). Since we expect our instrument to be sensitive to these
changes as they affect self-image, we will review briefly the current research
informing our validating hypotheses.
Contextual Factors Relevant to Self-Image
During Early Adolescence
Interpersonal interaction, present throughout the life span, is one con-
textual factor relevant to self-image. In addition, early adolescence is marked
by key changes in the social context, all potential influences on self-image
at this time.
First, most young adolescents change schools. Although some young
people in our country still attend K-12 (i.e., including grades kindergarten
through 12) or K-8 schools, most move from an elementary school to a mid-
dle or junior high school during these years. While any school change is
thought to be somewhat stressful (e.g., Blyth et al., 1983), the early adoles-
cent school change is especially likely to be so in that it typically involves
the change from a single teacher and classroom to multiple teachers and class-
es, more students per grade, and perhaps greater distance from home (Ham-
burg, 1974). Recent research has demonstrated clear effects on behaviors such
as smoking, drinking, dating, and violence of the specific grade levels present
in the school (Blyth et al., 1983). In addition, some evidence indicates that
this school change during early adolescence is particularly difficult for girls
who are simultaneously experiencing other changes (Simmons et aL, 1979).
Second, family interactions may change at this time, at least partly in
response to pubertal change and the symbolic meaning this has for impend-
ing adult status (Papini and Datan, 1983; Steinberg, 1981; Steinberg and
Hill, t978). Young adolescents and their parents need to begin the process
of reworking their relationships at this time to facilitate the increased need
for autonomy by the adolescent (Hill, 1980; Hill and Steinberg, 1976).
Third, peer relationships change by becoming more intimate, at least
for girls, and increasingly involving a component of sexual interest (Crock-
ett et al., 1984). The peer group becomes more structured and complex,
4. 96 Petersen, Schulenberg,Abramowitz,Offer,and Jarcho
with more groups including both boys and girls (Dunphy, 1963). Peers be-
come more important to young adolescents, although typically with no
decrease in the strength of the relationships with parents (Conger and Peter-
sen, 1984).
With changes occurring in these three social contexts, it is possible that
self-image related to these domains might change as well, either in the
degree of positive or negative valence or in the content of self-image in these
areas.
Ontogenetic Change in Self-lmage During Adolescence
Two major ontogenetic changes may affect self-image, particularly
during early adolescence. First, puberty involves major changes in appear-
ance as the child is transformed into an adult in terms of size, secondary
sex characteristics, and reproductive capacity (cf. Petersen and Taylor,
1980). Any of these features are likely to affect the young adolescent's self-
image as well as theway in which others view him or her.
A second major change is in cognitive capacity. Typically, during ear-
ly adolescence, the young person develops the capacity for abstract think-
ing, "thinking about thinking" (e.g., Elkind, 1974; Keating, 1980). This
enables young people to think about themselves-who they are and who
they would like to become.
Both the physical and cognitive changes during early adolescence
could lead to normative change or disruption in the self-image. The "classi-
cal" view of adolescent development, as distinct from the "empirical"
view--to use Coleman's (1978). definitions-proposes such normative
tumult. Derived from psychoanalytic theory, the classical view maintains
that adolescence is and must be a time of turmoil and emotional stress (e.g.,
Blos, 1962, 1979; Freud, 1958, 1969). Conversely, the empirical view, based
on findings obtained from representative samples of adolescents, holds that
adolescence is not disruptive for all, and is relatively calm for most or many
adolescents (e.g., Bandura, 1964; Douvan and Adelson, 1966; Offer and
Offer, 1975). The theoretical perspective proposing that all adolescents ex-
perience turmoil is therefore not supported by the data. At the same time,
some evidence indicates that the changes in social context as well as the on-
togenetic changes described earlier can be stressful, particularly when they
occur simultaneously. In addition, young people who are already psychologi-
cally vulnerable when they reach adolescence are likely to have greater
difficulty.
The study of self-image is central to this controversy, and hence it is
central to the understanding of adolescent development. Since self-image is
considered as an indicator of functioning and adjustment, the extent to
5. Measuring Self-Image of Young Adolescents 97
which adolescence is a difficult time ought to be reflected in self-image
scores. From the classical view of adolescent development, it would follow
that all adolescents experience some disruptions of self-image. Conversely,
the empirical view would assert that while some adolescents experience dis-
ruptions of self-image as a result of the cumulation of specific psychological
or contextual stressors, many or most do not (e.g., Abramowitz et at.,
1984).
Although the evidence is far from conclusive on the issue of self-image
development during adolescence (e.g., Wylie, 1979), there have been some
rather consistent findings. While our purpose is not to review comprehen-
sively this literature, it is useful to summarize the consistent patterns for
what they reveal about the construct of self-image. The most consistent
finding is that the content of self-image is continuous across the adolescent
years, as measured by factor structure invariance (e.g., Dusek and Flaherty,
1981; Kokenes, 1974; Michael et al., 1975; Monge, 1973). In addition, self-
image has been found to be fairly stable in terms of each"adolescent's posi-
tion relative to other adolescents over time (e.g., Carlson, 1965; Dusek and
Flaherthy, 1981; Engel, 1959; O'Malley and Bachman, 1983).
From absolute changes in self-image and self-esteem scores during
adolescence, it does appear that both self-image and self-esteem increase
over the adolescent years (Bachman et al., 1978; McCarthy and Hoge, 1982;
Offer and Howard, 1972; O'Malley and Bachman, 1983). Also, there is con-
siderable evidence to suggest that self-image and self-esteem become more
negative during early adolescence, compared to earlier and later periods
(e.g., Piers and Harris, 1964; Simmons and Rosenberg, 1975; Simmons et
al., 1973, 1979). Explanations for this decline have emphasized the stressors
characteristic of early adolescence, such as the entrance into junior high
school, the onset of puberty, the beginning of heterosexual concerns, and
concerns about peer-group acceptance (Hamburg, 1974; Rosenberg, 1979;
Simmons et al., 1979). For example, as Simmons and colleagues (Blyth et
al., 1983; Simmons et al., 1979; Simmons and Rosenberg, 1975) have
demonstrated, the entrance into junior high school, the onset of puberty,
and the beginning of dating were associated with declines in self-esteem
scores for girls, with girls experiencing all three stressors simultaneously be-
ing most susceptible to declines in self-esteem. Nevertheless, except for the
longitudinal research of Simmons and colleagues (in which a global measure
of self-esteem was employed), the only study indicating a drop in self-image
during early adolescence has utilized a cross-sectional design. Thus, conclu-
sions about this issue await further longitudinal studies.
A final consistent finding in studies of self-image during adolescence
concerns sex differences, with girls generally having poorer self-images than
boys during adolescence (Gove and Herb, 1974; Offer and Howard, 1972;
Offer et al., 1982; Simmons and Rosenberg, 1975). However, when self-
6. 98 Petersen, Schulenberg, Abramowitz, Offer, and Jarcho
image is measured multidimensionally, sex differences are found, but only
on certain dimensions of self-image. For example, both Dusek and Fiaherty
(1981) and Flaherty and Dusek (1980) found that males scored higher on
achievement/leadership and masculinity/femininity factors, and females
scored higher on the congeniality/sociabilityfactors. In addition, Petersen
(1981) found that with the OSIQ, while boys had higher overall self-image
scores than girls, boys had significantly higher scores on only 3 of the 11
scales (Emotional Tone, Body and Self-Image, and External Mastery), and
girls had significantly higher scores on 1 scale (Morals).
Hence, over the adolescence years, self-image appears to be a fairly
continuous construct (in terms of factor invariance), with subjects showing
stability of the construct over time relative to others. The extent to which
the self-image is positive may decline during early adolescence, but rises
steadily over middle and late adolescence. Also, while boys may have some-
what higher overall self-image than girls during adolescence, specific
aspects of self-image may show sex differences of varying magnitude and
direction.
METHODS
The SIQYA
As indicated, older and younger adolescents differ in cognitive and
pyschological functioning; therefore, self-image questionnaires developed
for older adolescents may sometimes be beyond the understanding of young
adolescents. Young adolescents may be more similar both cognitively and
psychologically to younger children, suggesting that a self-image measure
developed for children would be the most appropriate for young adoles-
cents. Because we may wish to follow self-image over the adolescent period,
however, a downward extension of instruments used with older adolescents
may be desirable. Since this was the case with our own research, we devel-
oped the SIQYA, which is a downward extension of the Offer Self-Image
Questionnaire (Offer et al., 1982).
The OSIQ, a self-descriptive measure of adjustment for adolescents
aged 14 to 18 years, has been used with thousands of adolescents in this
country and abroad (Offer et al., 1982). This questionnaire has been used
with a variety of subgroups and has been found to differentiate disturbed
adolescents (i.e., schizophrenics and neurotics) from those not seen clinical-
ly, as well as delinquent from nondelinquent adolescents (Offer et at.,
1982). The OSIQ contains 130 items and consists of 11 scales: (1) Impulse
Control, (2) Emotional Tone, (3) Body and Self-Image, (4) Social Attitudes,
7. Measuring Self-Image of Young Adolescents 99
(5) Morals, (6) Sexual Attitudes, (7) Family Relations, (8) External Mastery,
(9) Vocational and Educational Goals, (10) Psychopathology, and (11) Su-
perior Adjustment. The scales demonstrate sufficient psychometric proper-
ties, with good indices of internal consistency. All of the. scales, with the
exception of the Sexual Attitudes scale (which also failed to differentiate
among the various subgroups of adolescents) are significantly correlated.
Hence, although each scale taps a different aspect of self-image, the various
scales together tap a common dimension (Offer et al., 1982).
We revised some of the items troublesome to sixth-graders. The
Morals scale generally produced difficulties for young people of this age (as
represented by inadequate internal consistency coefficients), so we deleted
the entire scale. In addition, we deleted the Sexual Attitudes scale because
school personnel feared that it would be troublesome to young adolescents
(or their parents).
After revision, the SIQYA contains 98 items and consists of nine
scales; these include 72 original OSIQ items, 9 revised OSIQ items, and 17
new items. The Body and Self-Image scale was revised to focus only on
Body Image, a particularly relevant dimension for young adolescents. In
addition, 10 of the original or revised OSIQ items were moved to other
scales. To reflect the wording and content of the revised and new items,
three of the scales were renamed. The scales are described in Table I, by the
number of positively and negatively worded items per scale and a sample
item to illustrate the nature of each scale.
Responses to the SIQYA items are made with a 6-point Likert-type
scale, ranging from (1) "describes me very well," to (6) "does not describe
me at all." Half of the items are worded negatively, but items are scored so
that a high score.indicates high self-image.7
Samples
The SIQYA was developed as part of a large study of biopsychosocial
development related to mental health during early adolescence. The study
utilized a cohort-sequential longitudinal design (Baltes, 1968; Schaie, 1965).
The study sample was selected randomly from two successive sixth-grade
cohorts of two suburban school districts and was followed through the
eighth grade. The total study sample numbers 335, with about equal num-
ber of boys and girls in each cohort. Data collection for Cohorts I and II
began in 1978 and 1979, respectively, and the SIQYA was administered dur-
ing group testing sessions in the fall of each school year.
7Withthe OS1Q,a higherscoreindicatespoorerself-image.
8. 100 Petersen, Schulenberg, Abramowitz, Offer, and Jarcho
Table I. SIQYA Scale Descriptions and Reliabilities
Reliability
Number of Boys Girls
Scale (item example) items (direction) (N = 164) (N = I79)
1. Impulse Control 8 (4+/4-) 0.72 0.76
(I keep an even temper
most of the time)
2. Emotional Tone 11 (3+/8-) 0.81 0.85
(I feel nervous most
of the time)
3. Body Image II (5+/6-) 0.81 0.77
(1 feel proud of my body)
4. Peer Relationships l0 (4+/6- ) 0.85 0,81
(l think that other
people just do not
like me)
5. Family Relationships 17 (7 +/10-) 0,88 0.88
(My parents are usually
patient with me)
6. Mastery and Coping 10 (5+/5- ) 0.75 0.67
(I am fearful of
growing up)
7. Vocational-Educational
Goals 10 (6+/4 - ) 0.70 0.72
(1 enjoy learning new
things)
8. Psychopathology II (11 -) 0.79 0.80
(I fear something
constantly)
9. Superior Adjustment 10 (9+/1 -) 0.68 0.54
(I am a leader in school)
Only 6°70 of the participants dropped out, and 4°7o moved from the
area; a comparison of SIQYA scores of the adolescents who left the study
with scores of those who remained in the study revealed no differences in
this respect. The comparison of SIQYA scores in the study sample with
those of the rest of the unselected populations enabled us to conclude that
the students in the study sample were neither initially nor subsequently
different from their nonstudied peers. The study sample was almost entirely
White and from middle to upper middle class families. The sample was
above average in IQ (mean of 115.9 and 113.8 for the sixth and eighth
grades, respectively) and relatively homogeneous with respect to age (mean
age at January 1 of sixth grade was 11.60 years, SD = 0.34). Other demo-
graphic characteristics of the sample are more fully described elsewhere
(Richardson et al., 1984).
To demonstrate the psychometric properties of the SIQYA, three sam-
ples from this study were utilized, each to address a psychometric question.
9. Measuring Self-image of Young Adolescents 101
The first sample, described above, will be called the study sample (n = 335
students). The second sample, called the longitudinal sample, was identified
for the purpose of longitudinal analyses and includes students of the study
sample who were present for at least four of the six assessment sessions and
at least four of six interviews (n = 253 students), s The third sample, the
validating sample, consists of the Cohort II eighth-grade retest control
group (n = 343 students); this is the largest sample in which the completely
revised form of the SIQYA was administered.
RESULTS
Reliability
The reliabilities of the scales were measured by Cronbach's (1951) coeffi-
cient alpha for interitem consistency, which provides a good minimum esti-
mate of reliability (Lord and Novick, 1968). In the right-hand column of
Table I, the alpha coefficients obtained from the validating sample are
given for boys and girls. These reliabilities are well within the acceptable
range for instruments of this type, particularly considering that alpha is es-
pecially responsive to the number of items in the scale, which here is rela-
tively low.
Validity
The various domains of self-image appear to be adequately repre-
sented in the content of SIQYA items. The content of three scales (Peer
Relationships, Family Relationships, Vocational/Educational Goals) fo-
cuses on social contexts important in adolescence. The remaining scales fo-
cus on individual aspects of the self-image. In adapting the OSIQ, which
taps areas considered important to the psychological world of the adoles-
cent (Offer et al., 1982), our aim was to revise and write new items so that
each scale included at least 10 valid, internally consistent items representing
~Thisdefinitionof the longitudinalsamplemadeit possiblefor subjectsto be absent for one or
twoof the threeSIQYAsessions.Although71% of the subjectshad completedata, 27% missed
one sessionand only2070missedtwo sessions. For each subject in this sample with missing
data, missingscalevalueswereestimatedaccordingto a formulabasedon the subject'saver-
age deviationsfrom the group scalemeans during the times he or she was present, and the
group scalemeansduring the timehe or she was absent (seePetersenand Ebata, in prepara-
tion, for furtherinformation).Therewereno significantdifferencesin SIQYAscoresbetween
those with completedata and those with missingdata.
10. 102 Petersen, Schulenberg, Abramowitz, Offer, and Jarcho
that domain of self-image. We fell short of our numerical goal with the Im-
pulse Control scale, but its internal consistency is high.
Construct validity was ascertained in a number of ways. First, to
demonstrate criterion-related validity, SIQYA scale scores and total SIQYA
score (mean of scale scores) were compared with the Rosenberg Self-Esteem
Inventory (SEI; Rosenberg, 1965), a well-validated and accepted measure.
The SEI, a 10-item questionnaire, provides a global measure of self-esteem,
and has been found to be reliable and valid among young adolescents
(Rosenberg, 1965). The Rosenberg SEI and the SIQYA were administered
to the Cohort II seventh-graders (study sample) during the same testing ses-
sion. As is evident in Table II, all of the SIQYA scale scores and the total
SIQYA score are significantly and positively correlated with the SEI for
boys and girls, with the exception of Family Relationships for girls. The
magnitude of the correlations is typically high, and also somewhat larger
for boys (median r = 0.59) than girls (median r = 0.48). We might expect
the overall (average) self-image scores to be the one most similar conceptu-
ally to self-esteem (e.g., Rosenberg, 1965; Sherwood, 1965), an expecta-
tion supported by fairly high correlations in these data (0.72 and 0.62 for
boys and girls, respectively). In contrast, some evidence of discriminant va-
lidity may be seen in the lower correlations between self-esteem and impulse
control (0.37 and 0.34 for boys and girls, respectively). Therefore, associa-
tions between the SEI and SIQYA support the construct validity of the
SIQYA.
Construct validity was also evaluated by considering the extent to
which SIQYA scores could differentiate subjects reporting symptoms of
psychological maladjustment from those not reporting such symptoms. Us-
ing the OSIQ, Offer et al. (1982) found that disturbed and delinquent
Table 11.CorrelationsBetweenSIQYAScalesand the RosenbergSelf-Esteem
Inventory
SIQYA scale Boys (N = 54) Girls(N = 59)
1. ImpulseControl 0.35 0.34
2. Emotional Tone 0.73 0.48
3. BodyImage 0.54 0.28
4. Peer Relationships 0.38 0.44
5. FamilyRelationships 0.49 0.23~
6. Masteryand Coping 0.67 0.53
7. Vocational/Educational Goals 0.35 0.49
8. Psychopathology 0.59 0.35
9. SuperiorAdjustment 0.60 0.68
SIQYA Total score 0.73 0.58
aAll correlations are significantly different from zero except this one.
11. Measuring Self-Image of Young Adolescents 103
adolescents generally had lower self-image than normal adolescents; hence,
a similar finding would be expected with the SIQYA. In the study samples
in both cohorts, questions were asked during the eighth-grade spring inter-
views about the presence of symptoms of depression, anxiety, fears, and
school trouble (i.e., trouble-causing behavior in school)2 The specific items
are listed in the Appendix. A "yes" response to any of these questions was
considered to be a global indicator of the presence of the problem.
Young adolescents reporting each problem or not were compared in
multivariate analyses of variance on SIQYA scale scores. As Table III
shows, those reporting each problem had significantly poorer self-image
than the nonproblem group in the multivariate comparison and on the total
SIQYA score. Of the problem categories, Anxiety was least frequently re-
lated to SIQYA scores, with significant associations for three of nine scale
scores. The problems most frequently related to self-image, Depression and
School Trouble, were associated with all but two scale scores. Of the SIQYA
scales, Psychopathology was always significantly related to reports of
problems, as would be expected. Vocational/Educational Goals was the
scale least frequently related to problems, being associated only with reports
of School Trouble.
While the interview questions were only global indicators of psycho-
logical maladjustment, the SIQYA distinguished groups reporting these
problems from groups not reporting these problems. This provides further
evidence supporting the construct validity of the SIQYA, and is particularly
important in linking SIQYA scores with more serious mental health
problems.
Additional support for the construct validity of the SIQYA was found
in comparing the scores of adolescents from intact families with those from
families with marital problems in the eighth-grade longitudinal sample (co-
horts and sexes combined). During the eighth-grade parent interviews, par-
ents were asked questions pertaining to marital accord, discord, and divorce
or separation. Accordingly, the adolescents were placed in one of three
groups: intact (n = 182), discord (n = 15), or divorce/separation (n = 36).
There was a significant multivariate effect of marital status in a MANOVA
(F = 2.05, p < 0.007), with significant univariate effects on the Family
Relationships scale (F = 7.18, p < 0.0001). On this scale, adolescents in
the intact group had the highest mean (4.95), followed by the discord group
(4.58), with the divorce/separated group lowest (4.48). Therefore, the SIQYA
was sensitive to the presence of family problems.
9Aquestionwasalsoaskedaboutmanicbehavior,but thelowincidence(5% to 8%)of reports
makes the itemlessusefulin analysessuchas thosepresentedhere.
13. Measuring Self-Image of Young Adolescents 105
From the literature cited, sex differences in some aspects of self-image
would be anticipated; hence, consideration of sex differences is another means
of examining construct validity. In a MANOVA on the validating sam-
ple, a significant multivariate effect of sex was found (F = 3.88, p <
0.0001), with boys generally having higher scale scores. How,ever, at the
univariate level, the only significant sex difference was found with the Body
Image scale, where boys scored higher than girls (F = 15.02, p < 0.0001).
Finding overall differences such that boys report higher self-image is in ac-
cord with past research on sex differences of self-image in middle to late
adolescence (e.g., Gove and Herb, 1975; Offer and Howard, 1972; Offer
et al., 1982; Petersen, 1981; Simmons and Rosenberg, 1975). Furthermore,
the finding that boys scored higher than girls on the Body Image scale, a
scale that taps satisfaction with appearance, is consistent with research with
older adolescents (Lerner and Karabenick, 1974).
As indicated, self-image has been found to be continuous, in terms of
factor structure invariance, and stable, in terms of relative ranking of .in-
dividuals, over the adolescent years. Hence, similar findings with the SIQYA
would further indicate construct validity. To investigate factor structure
invariance across the grade levels, separate exploratory factor analyses
were performed on boys and girls (longitudinal sample). Scale scores at
each grade level and the obtained factor structures were compared with
coefficients of congruence (Tucker, 1950). The factor analysis procedures
and results are described elsewhere (Abramowitz, et al., in preparation);
here we focus on factor structure invariance. A two-factor solution was
found to be most appropriate in each case. Generally, the Emotional Tone,
Body Image, Peer Relationships, and Psychopathology scales had salient
factor loadings on the first factor, and the other five scales had salient fac-
tor loadings on the second factor. Table IV presents the coefficients of con-
gruence across the grade levels for boys and girls. As is evident, the
coefficients for boys are quite high, especially for adjacent grade levels; for
girls, the coefficients are somewhat lower. Overall, the coefficients indicate
a high degree of factor structure invariance in the SIQYA measure across
grades for both boys and girls. Also, the factor loading patterns of boys and
girls at each grade level are quite similar, as is evident in the coefficients on
the diagonal in Table IV. Hence, it appears that self-image scores obtained
from the SIQYA are continuous across the early adolescent years, a finding
consistent with previous research.
Stability of relative ranking of SIQYA scale scores over time was de-
termined by estimating stability coefficients on the scale scores of the lon-
gitudinal sample (cohorts and sexes combined) between the sixth and
seventh grade and between the seventh and eighth grade. The stability
coefficients were estimated according to formulae generated by Heise (1969),
14. 106 Petersen, Schulenberg, Abramowitz, Offer, and Jarcho
Table IV. Coefficients of Congruence for Two-Factor Solutions
of SIQYA Scale Scores Across Grade Levels for Boys and Girls"
Sixth Seventh Eighth
grade grade grade
Factor I 2 1 2 1 2
Sixth grade
1 0.98
2
Seventh grade
1 0.97
2
Eighth grade
1 0.95
2
0.94 0.90
O.99 0.94
0.98 0.97
0.97 0.96
0.84
0.93
0.99 0.95
0.93 0.99 0.87
"Coefficients below diagonal (diagonal is formed by italicized
numbers) pertain to boys; coefficients above diagonal pertain
to girls; and coefficients on diagonal pertain to the congruence
of boys and girls' factor structures.
which adjust year-to-year correlation coefficients for measurement error,
thus providing an estimate of true scores over time (O'Malley and Bachman,
1983). The stability coefficients were found to be generally high (at least
0.60 for one-year intervals and 0.44 for two-year intervals), and consistent
across scales. It is important to note that across the scales, stability coeffi-
cients were lower between the sixth and seventh grades than between the
seventh and eighth grades, suggesting that interindividual differences in in-
traindividual changes are less consistent between the sixth and seventh
grades (see Abramowitz et aL, in preparation, for further discussion). Over-
all, these rather high stability coefficients indicate that self-image, as mea-
sured by the SIQYA, is stable in terms of relative ranking, a finding that is
in accord with the research discussed.
The final assessment of the construct validity of the SIQYA involved
its sensitivity in detecting normative contextual changes, specifically the
transition into junior high school. As Simmons and colleagues (Simmons
et al., 1973; 1979) report, the transition to junior high school may have ad-
verse effects on self-esteem, especially for girls. For most of the students in
this study, the transition to junior high school occurred in the seventh grade
(see Schulenberg et al., 1984). To detect changes in SIQYA scale scores dur-
ing this time, a repeated measures MANOVA was performed on the lon-
gitudinal sample. Hypotheses about the nature of developmental trends
15. Measuring Self-Image of Young Adolescents 107
were tested by fitting a straight line (linear polynomial) and a curved line
(quadratic polynomial) to the data for the three grades. The results of these
analyses are reported fully elsewhere (Abramowitz et al., in preparation);
we will only summarize them here.
While a multivariate sex effect was found (in the direction reported
previously in this article), sex did not significantlyinteract with age, indicat-
ing that changes in self-image over time were similar for boys and girls.
Overall, self-image increased significantly from the sixth through eighth
grades, with greater increases between sixth and seventh grades than seventh
and eighth grades. An examination of the results by scale indicated that
scores on only two scales declined significantly over the grade levels: Body
Image and Superior Adjustment. The Emotional Tone, Peer Relationships,
Mastery and Coping, and Psychopathology scale means increased signifi-
cantly over the grade levels. The Family Relationships and Vocation-
al/Educational Goals scale means remained constant across the grade
levels, and the Impulse Control scale means followed a quadratic (but not
linear) trend and peaked in the seventh grade.
These results indicate that self-image, as measured by the SIQYA,
does not decrease as the students move into junior high school, a finding
discrepant from the research of Simmons and colleagues on self-esteem
changes. Nevertheless, the finding that the Body Image and Superior Ad-
justment scales, two scales that involve high degrees of social comparison
(see Abramowitz et al., 1984), decrease as the students enter junior high
suggests that the SIQYA does indeed detect transition-related changes in
self-image. Furthermore, as reported above, the SIQYA scale scores are less
stable, in terms of relative position, between the sixth and seventh grade
(the time when the transition occurred) than between the seventh and eighth
grade. This also suggests that the SIQYA is able to detect transition-related
effects of self-image.
DISCUSSION
This article presents data on an instrument (SIQYA) measuring
aspects of self-image among young adolescents. The nine scales show good
reliability assessed in terms of internal consistency. In addition, there is evi-
dence of the validity of the SIQYA assessed several ways. First, it is highly
correlated with a well-validated measure of self-esteem, the Rosenberg Self-
Esteem Inventory. Second, groups reporting four kinds of mental health
problems also report poorer self-image overall and poorer self-image on ap-
propriate scales. Third, young adolescents whose parents were divorced or
separated reported the poorest self-image (particularly in terms of family
16. 108 Petersen, Schulenberg, Abramowitz, Offer, and Jarcho
relationships), followed by the group whose parents reported marital dis-
cord, with young adolescents whose parents were married and did not
report discord expressing the most positive self-image. Fourth, overall sex
differences were found such that boys reported more positive self-images,
particularly in terms of body image.
The structure of self-image over time was found to be invariant among
young adolescents, suggesting that this structure is relatively stable over
these years. A similar picture was obtained by examining the stability coeffi-
cients over time. Self-image generally increased slightly over these years,
with the same pattern of change shown by boys and girls. The increases
were generally greater between the sixth and seventh than between the
seventh and eighth grades.
The results with the SIQYA are generally consistent with those ob-
tained in other research, except that we find little evidence of a decline in
most aspects of self-image during early adolescence. The scales that decline
are those most strongly linked to social comparison, a highly salient feature
of early adolescence.
In general, the results support the usefulness of a multiscale instru-
ment that provides information on the several important components of the
self-image during early adolescence. Although the pattern of results ob-
tained with the total self-image score is similar to that seen with self-esteem
measures, the results with specific scale scores (e.g., Body Image or Psy-
chopathology) demonstrate differentiated and informative patterns.
ACKNOWLEDGMENT
The authors are grateful to the entire staff of the "Early Adolescent
Study," we well as the young adolescents who participated with us in the
research.
APPENDIX
1. Depression: "Since you started sixth grade, have there been any periods
lasting two weeks or more when you felt sad, blue, depressed, or when
you lost all interest in things that you care about?"
2. Anxiety: "Since you started sixth grade, have you ever had a spell or at-
tack when all of a sudden, for no apparent reason, you felt frightened,
anxious, or very uneasy in a situation when most people wouldn't be
afraid?"
3. Fears: "Some people have such a strong fear of something or some situa-
tion that they try to avoid it even when there's no real danger. Since the
17. Measuring Self-Image of Young Adolescents 109
beginning of sixth grade, have you ever had such a fear of: (a) heights,
(b) being in a crowd, (c) going out of the house alone, (d) being in a
closed place, (e) being alone, (f) being near a harmless animal or a dan-
gerous one that couldn't get you?"
4. School Trouble: "Did you ever get into trouble with the principal or
teacher because of the way you acted in junior high school?"
REFERENCES
Abramowitz, R. H., Petersen, A. C., and Schulenberg, J. E. (1984). Changes in self-image
during early adolescence. In Offer, D., Ostrov, E., and Howard, K. I. (eds.), Patterns
of Adolescent Self-hnage, Jossey-Bass, San Francisco.
Abramowitz, R. H., Schulenberg, J. E., Petersen, A. C., and Offer, D. (in preparation). De-
velopmental change in self-image during early adolescence.
Bachman, J. G., O'Malley, P. M., and Johnson, J. J. (1978). Youth in Transition, Vol. 6: Adoles-
cence to Adulthood." A Study of Change and Stability in the Lives of Young Men, Insti-
tute for Social Research, Ann Arbor, Mich,
Baltes, P. B. (1968). Longitudinal and cross-sectional sequences in the study of age and genera-
tion effects. Hum. Dev. 11: 145-171.
Bandura, A. (1964). The stormy decade: Fact or fiction? PsychoL Schools 1: 224-231.
Blos, P. (t962). On Adolescence: A Psychoanalytic Interpretation, Free Press, New York.
Blos, P. (1979). The Adolescent Passage: Developmental Issues, International Universities Press,
New York.
Blyth, D. A., Simmons, R. G., and Carlton-Ford, S. (1983). The adjustment of early adoles-
cents to school transitions. J. Early Adoles. 3: 105-120.
Carlson, R. (1965). Stability and change in the adolescent's self-image. ChildDev. 36: 659-666.
Coleman, J. C. (1978). Current contradictions in adolescent theory. J. Youth Adoles. 7:1-11.
Conger, J. J., and Petersen, A. C. (1984). Adolescence and Youth: Psychological Develop-
ment hi a Changing World, 3rd ed., Harper & Row, New York.
Coombs, A. W. (1981). Some observations on self-concept theory and research. In Lynch, M.
D., Norem-Hebeisen, A. A., and Gergen K. J. (eds.), Self-Concept: Advances in The-
ory and Research. Ballinger, Cambridge, Mass.
Crockett, L., Losoff, M., and Petersen, A. C. (1984). Perceptions of peer group and friend-
ship in early adolescence. J. Early Adoles. 4: 155-181.
Cronbach, L. J. (1951). Coefficient alpha and the internal structure of tests. Psychometrika
15: 297-334.
Douvan, E., and Adelson, J. (1966). The Adolescent Experience, Wiley, New York.
Dusek, J. B., and Flaherty, J. F. (1981). The development of self-concept during the adoles-
cent years. SRCD Monogr. 46: 1-67.
Dunphy, D. C. (1963). The social structure of urban adolescednt peer groups. Sociometry, 26:
230-246.
Elkind, D. (1974). Children and Adolescents: Interpretive Essays on Jean Piaget, 2nd ed., Ox-
ford University Press, New York.
Engel, M. (1959). The stability of the self-concept in adolescence. J. Abnorm. Soc. PsychoL
58: 2t 1-215.
Flaherty, J. F., and Dusek, J. B. (1980). An investigation of the relationship between psycho-
logical androgyny and components of self-concept. 2'. Personal. Soc. PsychoL 38: 984-992.
Freud, A. (1958). Adolescence. Psychoan. Study Child 13: 255-278.
Freud, A. (1969). Adolescence as a developmental disturbance. In Caplan, G., and Lebovici,
S. (eds.), Adolescence: Psychosocial Perspectives, Basic Books, New York.
Gove, W. R., and Herb, T. R. (1974). Stress and mental illness among the young: A compari-
son of the sexes. Soc. Forces 53: 256-265.
18. 110 Petersen, Schulenberg, Abramowitz, Offer, and Jarcho
Hamburg, B. (1974). Early adolescence: A specific and stressful stage of the lifecycle. In Coel-
ho, G. V., Hamburg, D. A., and Adams, J. E. (eds.), Coping and Adaptation, Basic
Books, New York.
Heise, D. R. (1969). Separating reliability and stability in test-retest correlation. Am. Sociol.
Rev. 34: 93-101.
Hill, J. P. (1980). The family. In Johnson, M. (ed.), Toward Adolescence: TheMiddle School
Years, University of Chicago Press, Chicago.
Hill, J. P., and Steinberg, L. D. (1976). The development of autonomy during early adoles-
cence. In Jornadas sobre problemdticajuvenil, Fundaci6n Faustino Orvegozo Euzaguirra,
Madrid.
J6reskog, K. G., and S6rbom, D. (1978). EFAP-II: Exploratory Factor A nalysis Program: A
Fortran IV Program, National Educational Resources, Chicago.
Keating, D. P. 0980). Thinking processes in adolescence. In Adelson, J. (ed.), Handbook of
Adolescent Psychology, Wiley, New York.
Kokenes, B. (1974). Grade level differences in factors of self-esteem. Dev. Psycho/. 10: 954-958.
Lerner, R. M., and Karabenick, S. A. (1974). Physical attractiveness, body attitudes, and self-
concept in late adolescents. J. Youth Adoles. 3: 307-316.
Lord, F. M., and Novick, M. R. (1968). Statistical Theories of Mental Test Scores, Addison-
Wesley, Reading, Mass.
McCarthy, J. D., and Hoge, D. R. (1982). Analysis of age effects in longitudinal studies of
adolescent self-esteem. Dev. Psychol. 18: 372-379.
Mead, G. H. (1934). Mind, Self, and Society, University of Chicago Press, Chicago.
Michael, W. B., Smith, R. A., and Michael, J. J. (I975). The factorial validity of the Piers-
Harris Children's Self-Concept Scale for each of three samples of elementary, junior high,
and high school students in a large metropolitan school district. Educ. Psycho/. Meas.
35: 405-414.
Monge, R. H. (1973). Developmental trends in factors of adolescent self-concept. Dev. Psy-
chol. 8: 382-393.
Offer, D., and Howard, K. (1972). An empirical analysis of the Offer Self-Image Question-
naire for adolescents. Arch. Gen. Psychiat. 27: 529-533.
Offer, D., and Offer, J. B. (1975). From Teenage to Young Manhood, Basic Books, New York.
Offer, D., Ostrov, E., and Howard, K. I. (1981). The Adolescent: A Psychological Self-Portrait,
Basic Books, New York.
Offer, D., Ostrov, E., and Howard, K. I. (1982). The Offer Self-Image Questionnaire for Ado/es-
cents: A Manual, 3rd ed., Michael Reese Hospital, Chicago.
O'Malley, P. M., and Bachman, J. G. (1983). Self-esteem: Change and stability between ages
13 and 23. Dev. Psychol. 19: 257-268.
Papini, D. R., and Datan, N. (1983). Transition into adolescence: An interactionist perspec-
tive. Paper presented at the Biennial Meeting of the Society for Research in Child De-
velopment, Detroit, Michigan.
Petersen, A. C. (1981). The development of self-concept in adolescence. In Lynch, M. D., Norem-
Hebeisen, A. A., and Gergen, K. J. (eds.), Advances in Theory and Research, BaIlinger,
Cambridge, Mass.
Petersen, A. C., and Ebata, A. T. (in preparation). An estimation procedure for missing obser-
vations in longitudinal data.
Petersen, A. C., and Spiga, R. (t982). Adolescence and stress. In Goldberger, L., and Brez-
nitz, S. (eds.), Handbook of Stress; Theoretical and Clinical Aspects, Macmillan, New
York.
Petersen, A. C., and Taylor, B. (1980). The biological approach to adolescence, In Adelson,
J. (ed.), Handbook of Adolescent Psychology, Wiley, New York.
Piers, E. V., and Harris, D. B. (1964). Age and other correlates of self-concept in children.
J. Educ. Psychok 55: 91-95.
Richardson, R. A., Galambos, N. L., Schulenberg, J. E., and Petersen, A. C. (1984). Young
adolescents' perceptions of the family environment. J. Early Adoles. 4: 131-154.
Rosenberg, M. (1965). Society and the.Adolescent Self-Image, Princeton University Press, Prince-
ton, N.J.
19. Measuring Self-Image of Young Adolescents 111
Rosenberg, M. (1979). Conceiving the Self, Basic Books, New York.
Sameroff, A. J. (1977). Early influences on development, fact or fancy. In Chess, S., and Tho-
mas, A. (eds.), Annual Progress in Child Psychiatry and Child Development, Brun-
ner/Mazel, New York.
Schaie, K. W. (1965). A general model for the study of developmental problems. PsychoL Bull.
64: 92-107.
Schulenberg, J. E., Asp, C. E., and Petersen, A. C. (1984). School from the young adoles-
cent's perspective: A descriptive report. J. Early Adoles. 4: 107-130.
Sherwood, J. J. )1965). Self-identity and referent others. Sociometry 28: 66-81.
Simmons, R. G., and Rosenberg, F. (1975). Sex, sex-roles, and self-image. J. Youth Adoles.
4: 229-258.
Simmons, R. G., Rosenberg, F., and Rosenberg, M. (1973). Disturbance in the self-image at
adolescence. Am. SocioL Rev. 38: 553-568.
Simmons, R. G., Blyth, D. A., Van Cleave, E. F., and Bush, R. M. (1979). Entry into early
adolescence: The impact of school structure, puberty, and early dating. Am. SocioL Rev.
44: 948-967.
Steinberg, L. D. (1981). Transformations in family relations at puberty. Dev. PsychoL 17: 833-840.
Steinberg, L. D., and Hill, J. P. (1978). Patterns of family interaction as a function of age,
the onset of puberty, and formal thinking. Dev, PsychoL 14: 683-684.
Tucker, L. R. (1950). A method for synthesis of factor analysis studies. Personnel Research
Section Report, 984, Department of the Army, Washington, D.C.
Wells, L. E., and Marwell, G. (1976). Self-Esteem: Its Conceptualization and Measurement,
Vol. 20, Sage, Beverly Hills, Calif.
Wylie, R. C. (1974). The Self-Concept, Vol. 1, University of Nebraska Press, Lincoln.
Wylie, R. C. (1979). The Self-Concept, Vol. 2, Univeristy of Nebraska Press, Lincoln.