Psychological Assessment Copyrighl 1996 by the American Psychological Association, Inc.

1996. Vol. 8. No. 4, 369-377 1040-3590/96/$3.00

Social Competence and Behavior Evaluation in Children

Ages 3 to 6 Years: The Short Form (SCBE-30)
Peter J. LaFreniere Jean E. Dumas
University of Maine Purdue University

The factor structure and scale characteristics of the shortened versionof the Social Competence and
Behavior Evaluation Scale ( SCBE; P. J. LaFreniere & J. E. Dumas, 1995) are presented for a Quebec
sample and 3 U.S. samples, as well as age and gender differences in the prevalence of emotional and
behavioral problems and social competence throughout the preschool years. Principal-components
analyses identified 3 factors in all 4 samples: social competence (SC); anger-aggression (AA); and
anxiety-withdrawal (AW). Each 10-item scale was shown to have high interrater and test-retest
reliability, internal consistency, and temporal stability over a 6-month period. Across all samples,
boys were rated substantially higher on AA and lower on SC scales than girls, though no trends
toward a sex difference in AW were evident. A stepwise progression in SC with increasing age was
apparent in all samples. AA and AW behaviors showed a decline in the U.S., but not in the Qu6bec
sample.

The Social Competence and Behavior Evaluation Scale screening instrument to select samples of high-risk children;
(SCBE, formerly the Preschool Socio-Affective Profile) is an 80- (b) in longitudinal research to study the development of social
item Likert rating scale developed to assess patterns of social competence; (c) in intervention studies as an outcome measure
competence, emotion regulation and expression, and adjust- to assess treatment effects; and (d) in experimental research ad-
ment difficulties in children ages 30 to 78 months (LaFreniere, dressing various issues in social and emotional development.
Dumas, Capuano, & Dubeau, 1992). The questionnaire, typi- For a recent summary of this work, see LaFreniere and Dumas
cally completed by preschool teachers, is composed of eight (1995). In addition to research, educators and clinicians have
scales balanced for positive (competence) and negative used the SCBE to assess the degree and specific features of be-
(emotional or behavioral problems) items, covering an exten- havioral and emotional problems in children in public and
sive array of behaviors commonly seen within a preschool set- private day care and preschool centers, including Head Start
ting. As many emotional and behavioral problems in early programs.
childhood are rarely directly observed by psychiatrists, psychol- Because the SCBE is now being used for multiple purposes
ogists, or other mental health professionals, it is important to by diverse groups of professionals, we believe that a shortened
obtain reliable and valid information from people who do ob- version of the instrument that retains as many characteristics
serve young children under conditions in which such problems of the original as possible is desirable, because it would enable
typically occur. Once standardized, this information can pro- researchers and educators to function more efficiently given the
vide an important complement to other sources of infor- time constraints and other demands they face daily. The short
mation concerning the quality oflhe child's adaptation in other
form of the SCBE is intended to preserve several important
settings. characteristics of the original while reducing the overall length
The SCBE has been widely used over the past 8 years in re-
of the instrument from 80 to 30 items. These characteristics
search, educational, and clinical settings in the United States,
include the three-factor structure; the high reliability, internal
Canada, and Europe. Researchers have used the SCBE (a) as a
consistency, and temporal stability of the scales; and the relative
orthogonality of the two factors representing internalizing and
externalizing behavior profiles.
Peter J. LaFreniere. Department of Psychology, Universityof Maine: We approached the task of constructing the original SCBE
Jean E. Dumas, Department of Psychological Sciences, Purdue from a developmental-adaptational perspective, emphasizing
University. the functional significance of affect in regulating social ex-
The development of this short form of the Social Competence and change (Bowlby, 1980; Ekman, 1984; Izard, 1977, 1991; Plut-
Behavior Evaluation involved many people over the past 6 years. We chik, 1980, 1991; Sroufe, 1996). From this perspective emo-
would like to thank especially our research assistants at the University tions signal needs, attitudes, anticipations, and impulses toward
of Montreal--FranceCapuano and Diane Dubeau--who were essential
action. Discrete, primary emotions such asjo); love, anger, fear,
in recruiting a large sample for the initial normalization and validation.
and sadness are evident in infancy and generally shared across
Finally, we express our gratitude to several hundred preschool teachers,
without whom this work would have been impossible. cultures. They are widely recognized, not just as common ver-
Correspondence concerning this article should be addressed to Peter bal labels for emotional experience but as core species-specific
LaFreniere, Department of Psychology, Universityof Maine, 5742 Clar- motivational systems that shape or organize both behavior and
ence Cook Little Hall, Orono, Maine 04469-5742. development across the life span (Fischer, Shaver, & Carnochan,
369
 370 LAFREN1ERE AND DUMAS

1990: Sroufe, 1996; Sroufe, Schork, Motti, Lawroski, & La- scales in terms of etiology, temporal stability, or external corre-
Freniere, 1984). lates may be attributed to the initial differences in the psycho-
It is apparent from observational research in children's pre- metric characteristics of the scales rather than to substantive
school groups that affective expression and emotion regulation differences in the phenomenon under investigation. In contrast,
serve important relationship functions. Characteristic patterns the SCBE-30 is designed to produce a measure of internalizing
of emotion provide an important cue for observers interested in behavior that is as robust psychometrically as its externalizing
the ongoing adjustment of the child to the milieu. For example, counterpart, and relatively orthogonal to it.
in a recent study considerable differences were found in pre- Our clinical objective is to provide an instrument capable of
schoolers' ability to modulate negative affect arising from frus- describing behavioral tendencies for the purpose of socializa-
tration at losing a competitive encounter with a peer tion and education. One consequence of this orientation is our
( LaFreniere, 1996 ). Although many of these children were able emphasis on assessing the quality of the child's adaptation
to maintain a positive interaction with their partner, two dis- rather than sampling only the child's problem behaviors. The
tinct patterns of maladaptive behavior were identified. Children inclusion of a reliable, valid, and developmentally sensitive in-
who were rated by teachers as angry-aggressive expressed more dex of social competence allows teachers to orient classroom
negative affect after having lost a competition and were less able intervention efforts toward the child's strengths as well as vul-
to share a limited resource with the same peer in a subsequent nerabilities. Second, this emphasis allows for more refined de-
encounter. It appeared that these children were unable to regu- scription of developmental deviance by focusing on the pres-
late a moderate level of negative affect. Their concern with the ence or absence of positive behavior as well as problem behavior.
results of the competition led to more conflicts of interest, in- Finally, this focus may also enhance prediction of later disorder
creased competitiveness, and interpersonal conflicts with their because early investigators have reported better prediction from
partner. In contrast, children rated as anxious-withdrawn indexes of competence and ego maturity rather than from the
seemed to disconnect from their partner following the competi- absence of problems and symptoms (Kohlberg, LaCrosse. &
tion. Their affective expression, although not more negative, Ricks, 1972).
was not contingent on their partner's behavior or affect. Re- In this study we report the factor structure and scale charac-
search on the developmental processes underlying emotional teristics of the shortened version of the SCBE as well as age and
competence is just beginning to gather momentum; its ultimate gender differences in the prevalence of emotional and behav-
success will depend on how successfully researchers meet the ioral problems and social competence throughout the preschool
methodological challenges that have long hindered progress in years.
this area. The development of the SCBE-30 is one attempt to
meet this challenge. Method
Like its predecessor, the SCBE-30 is designed to capture this
affective quality of the child's relationships with teachers and Participants
peers. Specifically, we aim to provide (a) a standardized descrip- The database consisted of four distinct samples: three recruited in
tion of affect and behavior in context that is reliable, valid, and regions of the United States representing the Midwest (Indiana). West
useful for preschool teachers; (b) an efficient screening instru- (Colorado), and Northeast (Maine), and the original French Canadian
ment that differentiates specific types of behavioral-emotional sample from Montreal, Qu6bec, for a total of 2,646 participants. See
problems: (c) a developmentally appropriate assessment of Table 1. The Quebec sample consisted of 910 preschool children re-
children's positive social adaptation or competence; (d) mea- cruited from 80 different preschool classrooms from the Montr6al met-
sures of high and approximately equivalent internal consis- ropolitan area. The Indiana sample consisted of 854 preschool children
tency, reliability, and stability; and (e) measures sensitive to be- recruited from 50 different preschool classrooms from the Indianapolis
havioral change over time for the purpose of program evalua- and Lafayette areas; the Colorado sample of 439 preschool children was
drawn from 30 classrooms from the Boulder and Denver areas: and the
tion or prospective-longitudinal research.
Maine sample of 443 preschool children was recruited from 20 class-
A major impetus for our research program is to develop an rooms from the Bangor and Orono areas. For analytical purposes, age
instrument capable of addressing an important issue in pro- was treated as a categorical variable by classifyingparticipants into four
spective longitudinal research of children at risk for later disor- groups: 3 year olds ( 30 to 42 months): 4 year olds" ( 43 to 54 months ): 5
ders. As research goals become more precise and investigators ),ear olds ( 55 to 66 months); and 6 year olds ( 67 to 78 months ). A child
seek to understand both differential etiologies and sequelae of was thus designated, for example, a 3 year old if he or she was within 6
developmental disorders, greater discriminant validity is being
demanded of measures used to evaluate various problem be-
haviors across various ages. Scales produced through tradi- Table 1
tional factor analytic techniques inevitably yield variables com- Gender and Ages otParticipants in Four Samples
posed of unequal numbers of items, and consequently, quite
different psychometric characteristics. Typically, a large first Gender Age (years)
factor representing conduct disorder is found to be more in-
Sample Girls Boys N 3 4 5 6
ternally consistent, reliable, and stable than a much smaller fac-
tor representing emotional disorder or anxiety (e.g., Behar. Quebec 415 495 910 251 350 309
1977; Behar & Stringfield, 1974; Hinshaw, Morrison, Carte, & Indiana 440 414 854 44 160 404 246
Cornsweet, 1987; Hogan, Quay, Vaughn, & Shapiro, 1989). Colorado 210 229 439 46 155 172 66
Subsequent comparisons demonstrating differences between Maine 213 230 443 88 190 126 39
 SCBE-30 371

months of exactly 3 years of age at the time of the assessment. In the identical 10-item factors accounting for 52.9% and 49.8% of the
American sample, the four most populous ethnic groups in the United variance, respectively. Each factor was composed of conceptu-
States--White ( 72% ), Black ( 18% ), Hispanic ( 7% ), and Asian Ameri- ally homogenous items tapping affective characteristics (e.g.,
can (3%)--are well represented, with minority groups approximately "irritable," "gets mad easily") and social behaviors involving
equal to the 1993 national norms provided in the Statistical Abstract of
peers (e.g., "bullies younger children") or adults (e.g., "defiant
the United States (U.S. Bureau of the Census, 1991 ).
when r e p r i m a n d e d " ) . The first factor to emerge in each was
Teacher Evaluations labeled Anger-Aggression (AA), and it accounted for 29.2% and
26.7% of the variance, respectively; the second factor, Social
All children were evaluated by one or two classroom teachers at the
end of the fall session or during the spring session using the SCBE, which
Competence (SC), accounted for an additional 16.9% and
has a Likert-type 6-point response format: never' rarely:sometimes, of- 15.2% of the variance, respectively; and the third factor, Anxi-
ten, frequent(v, and always. All children were evaluated whose parents ety- Withdrawal (AW), explained an additional 6.8% and 7.9%
consented to their participation (approximately 70-80% across all of the variance, respectively. Subsequent analyses confirmed the
samples) and who had been enrolled in the day care or preschool for at presence of the same three factors in all U.S. samples, account-
least 2 months. ing for 54.0% o f the total variance for the Indiana sample, 52.4%
in the Colorado sample, and 53.1% in the Maine sample. (See
Construction o f the S C B E - 3 0 Table 2 for factor loadings for all four samples).
As described in the original validation study or the SCBE-80
(formerly PSP; LaFreniere et al., 1992), a principal-components anal-
Reliability, Consistency, and Stability
ysis using varimax rotation was pertbrmed on 16 item clusters. The
original 80 items were conceptually organized into bipolar item clusters Interrater reliability provides an assessment of the degree of
to assess patterns of emotion regulation (joyful-depressive, secure-anx- equivalence between different teachers' independent evalua-
ious, and tolerant-angry); peer relations (integrated-isolated, calm-
tions of the same child at the same point in time. Agreement
aggressive, prosocial-egotistical); and teacher-child relations (co-
operative-oppositional, autonomous-dependent). This analysis yielded was calculated for each of the three factor-scales (hereinafter
three factors accounting for 67.1% of the total variance. The first factor, scales) using Spearman-Brown estimates based on correlations
designated Social Competence, comprised the eight positive item clus- averaged across the Quebec, Indiana, and Maine sites. Because
ters (joyful, secure, tolerant, socially integrated, calm, prosocial, coop- only one teacher had completed the SCBE in the Colorado sam-
erative, autonomous); the second factor, labeled Externalizing Behav- ple, interrater agreement could not be estimated. As shown in
ior, comprised four negative item clusters (angry, aggressive, egotistical, Table 3, the reliability estimates were uniformly high across
oppositional); and the third factor, Internalizing Behavior, comprised samples, ranging from .78 to .91, This is quite comparable to
the remaining four negative item clusters (depressed, anxious, isolated, the reliability estimates for the SCBE-80, which ranged from
dependent). Subsequent analyses confirmed the presence of the same .72 to .89 (LaFreniere et al., 1992).
three factors in all U.S. samples, accounting for 77.6% of the total vari-
Internal consistency is an assessment of the degree to which
ance in the Indiana sample, 76.2% in the Colorado sample, and 75.7%
in the Maine sample ( LaFreniere & Dumas, 1995 ). the component items of a scale coalesce around a coherent and
This robust three-factor solution provided the starting point for this stable central tendency. The most widely used index of internal
study. In the first step, three criteria were used to reduce these three consistency is Cronbach's coefficient alpha, which represents
original factors from the 80-item instrument to three scales composed the average of all possible split-half reliabilities that can be per-
of 15 items each: level of endorsement, interrater reliability, and in- formed on a scale. Alpha statistics for all four samples are pre-
ternal consistency. In the second step, these 45 items were factor ana- sented in Table 4. The results show that all three scales have a
lyzed and reduced further by selecting the 10 items with the highest high degree of internal consistency in each sample, with scores
factor Ioadings within each of the three factors. This new 30-item set ranging from .80 to .92. Again, this is quite comparable to
was then analyzed, using a principal-components analysis with varimax the alphas for the SCBE-80, which ranged from .79 to .91
rotation. Given the favorable participants-to-variables ratio (over 30:
(LaFreniere et al., 1992 ).
1), we split the Quebec sample in half, using an odd-even procedure,
and conducted two additional factor analyses to verify the stability Data on test-retest reliability and temporal stability are
of the factor structure. Subsequently. each of the three American sam- based on subsamples of 29 French Canadian children (2
ples were analyzed separately using the same principal-components teachers), 409 Indiana children (16 teachers), and 45 Maine
analysis. children (2 teachers) selected at random and reevaluated by
In addition to examining the factor structure of the instrument, the their teachers 2 weeks (Quebec) or 6 months (Indiana, Maine)
scales were examined for internal consistency using Cronbach's alpha, after the initial evaluation. As shown in Table 5, Pearson corre-
interrater reliability, test-retest reliability, and temporal stability. Age lations across a 2-week interval were uniformly high ( . 7 8 - . 8 6 )
and gender differences were also examined using analysis of variance and comparable to the 2-week reliability estimates for the
(ANOVA). These analyses were performed for the four samples wher-
SCBE-80, which ranged from .74 to .87 (LaFreniere et al.,
ever the data permitted. However, because of uneven cell sizes in the
1992). As would be expected, slightly lower correlations were
individual American samples, the ANOVA was performed on a single
U.S. sample consisting of all three American samples combined. obtained when participants were reevaluated across a 6-month
interval (.75-.79 for in the Indiana sample during the same ac-
Results ademic year: .61-.69 in the Maine sample in different academic
years).
Factor Analyses
Intercorrelations among the scales demonstrated relative or-
Split-half analyses for the Quebec sample revealed high sta- thogonality between the AA and the AW scales in all four sam-
bility across factor structures. Each analysis generated three ples (r = .02,. 16, .29, .02), whereas the SC scale was signifi-
372 LAFRENIERE AND DUMAS

Table 2
Factor Loadings of SCBE-30 Items for Four Samples
Original SCBE
SCBE-30 Quebec Indiana Colorado Maine
item no. Item no. Subscale (N = 910) (N = 854) (N = 439) (N = 443)

Anger-Aggression
1 10 Angry .84 .80 .82 .85
2 7 Angry .78 .62 .63 .71
3 74 Oppositional .77 .79 .78 .81
4 44 Aggressive .76 ,77 .76 .82
5 30 Angry .76 .71 .67 .80
6 8 Oppositional .71 ,79 .75 .79
7 42 Aggressive .71 ,74 .72 .73
8 62 Oppositional .66 ,62 .63 .70
9 32 Aggressive .65 .77 .71 .72
10 71 Oppositional .58 .72 .69 .67

Social Competence
1 35 Calm .73 .67 .53 .76
2 45 Prosocial .72 .70 .72 .76
3 39 Calm .69 .66 .74 .80
4 58 Integrated .68 .60 .66 .60
5 43 Prosocial .66 .61 .73 .64
6 17 Joyful .65 .45 .60 .49
7 66 Cooperative .65 .49 .60 .60
8 50 Calm .63 .55 .68 .59
9 60 Prosocial .58 .63 .58 .63
10 63 Cooperative .57 .53 .65 .60

Anxiety-Withdrawal
1 36 Isolated .70 .78 .70 .72
2 34 Isolated .69 .80 .60 .80
3 24 Depressive .67 .67 .67 .62
4 28 Anxious .66 .75 .69 .78
5 49 Isolated .64 .66 .51 .69
6 23 Anxious .62 .74 .75 .75
7 57 Isolated .54 .67 .57 .65
8 1I Anxious .50 .50 .57 .66
9 6 Depressive .49 .53 .35 .45
10 2 Depressive .49 .63 .45 .62

Note. SCBE-30 = Social Competence and Behavior Evaluation Scale: Short Form.

cantly and negatively correlated with b o t h AA (r = - . 3 7 , - . 5 7 , lier. Given the uneven distribution o f age groups in the three
- . 5 8 , - . 5 4 ) and AW ( r = - . 3 8 , - . 4 2 , - . 4 3 , - , 3 0 ) in all four A m e r i c a n samples (see Table 1 ), they were c o m b i n e d to form
samples. a single sample with m o r e substantial cell sizes for the 3- and 6
year old groups. As shown in Figures 1 and 2, the main effect
Age and Gender Effects for age for the SC scale was significant in b o t h the French Cana-
dian, F ( 2, 905 ) = 17.78, p < .001, and the c o m b i n e d A m e r i c a n
To assess age and gender differences, 2 ( G e n d e r ) x 4 (Age)
ANOVAs were c o n d u c t e d using the three scales described ear-
Table 4
Internal Consistency of SCBE-30 Scalesfor Four Samples
Table 3
Interrater Agreement of SCBE-30 Scalesfor Three Samples Cronbach's alpha

Scale Qu6bec Indiana Maine Scale Quebec Indiana Colorado Maine

Anger-Aggression .83 .91 .79 Anger-Aggression .92 .92 .92 .92

Social Competence .85 .83 .87 Social Competence .90 .86 .90 .88
Anxiety-Withdrawal .86 .78 .88 Anxiety-Withdrawal .85 .77 .81 .84

Note. SCBE-30 = Social Competence and Behavior Evaluation Scale: Note. SCBE-30 - Social Competence and Behavior Evaluation Scale:
Short Form. Short Form.
 SCBE-30 373

Table 5 portion of the Indiana sample (n = 517 ), we were able to obtain

Test-Retest Reliability and Temporal Stability teacher ratings of conduct disorder and anxiety-withdrawal
of SCBE-30 Scales measures derived from the Revised Behavior Problem Checklist
(RBPC; Hogan et al., 1989). The Pearson correlation was .67
Temporal stabilityb for the two measures of anxiety-withdrawal and .87 for the re-
Test-retesta
Scale (Quebec) Indiana Maine lation between anger-aggression (SCBE-30) and conduct dis-
order (RBPC). Each of these correlations parallels earlier anal-
Anger-Aggression .86 .79 .69 yses (LaFreniere et al., 1992) between the SCBE-80 and the
Social Competence .82 .78 .61 Child Behavior Checklist (CBCL-Teacher Report Form;
Anxiety-Withdrawal .78 .75 .63 Achenbach & Edelbrock, 1981 ) and confirms our expectation
Note. SCBE-30 = Social Competence and Behavior Evaluation Scale: of moderate associations, high enough to provide convergent
Short Form. validity but also different enough to justify a new measure.
a 2-week interval, b 6-month interval.
Discussion

samples, F(3, 1729 ) = 41.75, p < .001, with teachers reporting Principal-components analyses clearly identified three fac-
greater social competence in older participants. One-way tors in all four samples representing social competence, anger-
ANOVAs, using Duncan's correction for multiple comparisons, aggression, and anxiety-withdrawal, respectively. Each factor-
revealed a stepwise progression in social competence for age scale, composed of 10 items, was shown to have high interrater
groups for both sexes in both samples (3 < 4 < 5 < 6; p < and test-retest reliability, internal consistency, and temporal
.05). Age effects in problem behaviors were evident in the U.S. stability over a 6-month period. It is noteworthy that each scale
sample but not in the Quebec sample. A stepwise progression was virtually synonymous with the corresponding scale from
in the decline of angry-aggressive behavior was found in the the original SCBE, despite the significant reduction from 80 to
U.S. sample for boys, whereas U.S. girls showed a decline in AA 30 items. This finding permits confidence in applying the exten-
behavior after age 4 (5, 6 < 3, 4; p < .05). In addition, an age sive validation studies of the original (LaFreniere et al., 1992)
effect was significant in the U.S. sample for anxious-withdrawn to the new, abbreviated version.
behavior, F ( 3, 1729) = 13.16, p < .001. Duncan tests revealed The Social Competence scale taps a broad range of behaviors
an overall decline in teacher reports of AW after age 4 (5, 6 < 3, designed to assess the positive qualities of the child's adaptation
4; p < .05 ) for gifts and a similar effect for boys, although no rather than specific behavioral competencies. As a developmen-
difference was found between 4 and 5 year old boys). In con- tal construct, social competence refers to behaviors that indi-
trast, no significant main effects for age were found in either AA cate a well-adjusted, flexible, emotionally mature, and generally
or AW in the Quebec sample. prosocial pattern of social adaptation. In previous validation
There were main effects for gender for the S C - - F ( 1 , 9 0 5 ) = work with the SCBE-80 (LaFreniere et al., 1992), the socially
51.07, p < .001 (Quebec); F ( I , 1729) = 43.92, p < .001 competent group received the most positive nominations and
( U . S . ) - - a n d AA s c a l e s - - F ( 1 , 905) = 26.18, p < .001 the least negative nominations from peers, and it was highest in
( Quebec ); F ( 1, 1729 ) = 53.01, p < .001 (U.S.) - - f o r both sam- sociometric status. Because of its qualitative dimension, socio-
ples, with teachers assessing girls as more socially competent metric status is one of the most robust correlates of teacher rat-
and less aggressive than boys. In addition, there was a significant ings of social competence in preschoolers. Other research has
Age × Gender interaction for the AA scale in the Quebec sam- shown that socially competent preschoolers score high on as-
ple, F(2, 905 ) = 3.24, p = .040, but not in the overall American sessments of ego resiliency, express high levels of positive affect
sample. French Canadian boys and girls were not initially and relatively low levels of negative affect, and are well liked and
different at 3 years on that scale; however, older boys were rated appreciated by their teachers as well as by peers (LaFreniere &
as significantly more aggressive than older girls. In contrast to Sroufe, 1985; Sroufe et al., 1984; Waters & Sroufe, 1983).
the foregoing pattern of results, there were no significant gender The second scale, Anger-Aggression, may be considered a
differences in teacher evaluations on the AW scale in either sam- preschool equivalent of previous broadband syndromes labeled
ple. The means and standard deviations for both the United variously as conduct disorder (Peterson, 1961; Quay, 1983 ) or
States and Quebec samples are presented in Appendixes A and externalizing symptoms (Achenbach & Edelbrock, 1981 ). The
B, respectively. items on this scale describe angry, aggressive, selfish, and oppo-
sitional behaviors. It is interesting that emotion regulation and
affective expression were shown to be as central as aggressive
Construct Validity
behavior to this scale. Indeed, the item with the highest factor
Because of the extensive construct validation of the original loading in all four samples was "irritable, gets mad easily," and
80-item SCBE (Dumas & LaFreniere, 1993; Dumas, LaFren- poor frustration tolerance and defiance were among the other
iere, & Serketich, 1995; LaFreniere et al., 1992; LaFreniere & high loadings. Children with extreme scores on this scale tend
Dumas, 1992), Pearson correlations were computed between to express their negative emotions in ways that hurt or at least
these latter three indexes and their corresponding 10-item scales disturb others. They function poorly in social situations, where
from the SCBE-30. As expected, these correlations were ex- they require almost constant supervision, often to protect oth-
tremely high ( .92-.97 ), reflecting the high degree of conceptual ers from their outbursts. The angry-aggressive group "was pre-
unity retained in the short version. In addition, for a large pro- viously shown to be the most interactive with peers but also the
374 LAFRENIERE AND DUMAS

most rejected--results that correspond precisely with theoreti-

cal expectations because aggression is a well-known behavioral
correlate of peer rejection across a wide age span (Asher &
Dodge, 1986; Coie, Dodge, & Kupersmidt, 1990; LaFreniere &
Sroufe, 1985).
The third scale, Anxiety-Withdrawal, is composed of items
describing anxious, depressed, isolated, and overly dependent
behavior. It is similar to previous broadband indexes labeled
affective disorder or internalizing symptoms, although somatic
complaints, eating and sleeping irregularities, and similar prob-
lems were not included in the item list because of the preschool
context of the evaluation, In previous validation research, the
anxious-withdrawn group was observed to be significantly less
interactive than all other groups, although not necessarily ne-
glected or rejected by their peers. However, AW preschoolers did
spend much of their time in activities on the periphery of group
life, such as onlooking and parallel play, and they were also
prone to be alone and unoccupied. In keeping with the pattern
of affective and behavioral signs of maladjustment, children
with extreme scores on this scale are seen by their teachers as
sad, depressed, tired, and worried, as well as isolated. They show
little interest in the activities that their peers typically delight in.
These children tend to have poor self-concepts and show high
levels of immaturity, as they often seek adult attention in situa-
tions that do not require it and find it difficult to perform tasks
within their capabilities without regular assurance, giving up
easily when others would persist.
This basic division between these two broad types of disorder
has been extensively verified in early and middle childhood as
well as in adolescence (Achenbach & Edelbrock, 1981; Behar &
Stringfield, 1974; Kohn & Rosman, 1972; Peterson, 1961:
Quay, 1983). However, unlike broadband assessments of exter-
nalizing and internalizing problems, the AA and AW scales of
the SCBE-30 are conceptually distinct, and each is defined in
terms of affective and behavioral cues. Empirically, we have
shown that these scales are much more orthogonal than the
broadband measures of internalizing and externalizing behav-
ior typically used in developmental research. We view this as a
significant improvement because it affords the opportunity for
more effective longitudinal research regarding differential etio-
logies and sequelae of these developmental disorders.
Given the extensive samples of children evaluated in this
study, these data are particularly robust with respect to age and
gender differences. Across all samples, boys were shown to be
rated substantially higher on measures of anger-aggression and
lower on social competence than girls, although no sample
showed even a trend toward a sex difference in anxiety-with-
drawal. Taken together, these results would appear to indicate
that boys are, on the whole, not as well adapted to the preschool
environment as girls, at least from the teacher's standpoint. It
should be noted that over 95% of the teachers, from the several
hundred programs that provided these data, were female, a pro-
portion that is representative of preschool programs in North
America. This introduces two potential sources of gender bias.
A female bias in the ratings of the children's behavior may be
present, and a bias may be expressed in the construction of an
Figure 1. Age and gender differences in SCBE-30 scales for the U.S. environment that is more suitable to preschool girls than pre-
sample. SCBE-30 = Social Competence and Behavior Evaluation school boys. It is possible that the preschool environment is less
Scale: Short Form appealing and more restrictive for a higher percentage of boys
SCBE-30 375

than girls. It is also likely that a greater proportion of these chil-

dren respond to these conditions with externalizing behaviors.
If both of these propositions were true, then we would expect
a much higher percentage of disruptive behavior problems in
b o y s - - a fact that is easily substantiated but rarely explained
(Dumas, 1994; Richman, Stevenson, & Graham, 1982; Rutter
& Garmezy, 1983). It may also be argued that boys are geneti-
cally predisposed toward higher levels of aggressive or disruptive
behavior; however, in our view of adaptation, the environment
must also be part of the equation.
Regarding age differences, developmental trends toward a
stepwise progression in social competence for boys and girls was
clearly evident in both samples, in line with theoretical expec-
tations. Similar to the validation criterion for IQ items, we view
a clear linear trend of increasing social competence with age as a
basic component of the validation process of this developmental
construct.
It is unclear whether cultural or linguistic differences underly
differences between the U.S. and Qu6bec samples regarding age
trends in behavior problems. Angry-aggressive behaviors
showed a clear stepwise decline (especially in boys) for the com-
bined U.S. sample but no main effect for age in the Qu6bec
sample. Similarly, the U.S. data reveal a less pronounced but
significant drop in anxious-withdrawn behavior among older
preschoolers, with no apparent age trend in the Quebec data.
No definitive interpretations of these differences in age trends
are possible at this point, especially because no differences were
predicted and the age range of the Quebec sample was narrower
than that of the U.S. sample. Currently, large samples are being
tested with French and Spanish versions of the SCBE in Europe
and North America, which will allow more definite conclusions
than any single 2-point comparison.
In conclusion, we have shown that reducing the SCBE-80
to an abbreviated 30-item instrument does not result in any
detectable loss in interrater agreement, test-retest reliability, in-
ternal consistency, or temporal stability. In addition, the SCBE-
30 retains the original three-factor structure. Because of the re-
search demands of economy, data reduction, and discriminant
validity, it will often be advantageous to use this new short form,
particularly when teachers are asked to assess moderate-to-large
numbers of children or are required to make multiple assess-
ments of different aspects of the same children. In clinical ap-
plication, the SCBE-30 would be ideal as part of a screening
assessment or outcome assessment, again when large samples
are used.
However, for most clinical purposes, we recommend the
SCBE-80, not only because of the more detailed description
provided by the profile of the eight basic scales, but also because
of the nationally standardized norms and case studies provided
in the manual (LaFreniere & Dumas, 1995). T scores and per-
centile scores provide the essential context for understanding a
target child's relative standing with respect to all children of the
same gender in the standardization sample. T scores are based

Figure 2. Age and gender differences in SCBE-30 scales for the Que-
bec sample. SCBE-30 = Social Competence and Behavior Evaluation
Scale: Short Form.
376 LAFREN1ERE AND DUMAS

on the cumulative probabilities in the raw score distribution Replication and validation within a kindergarten sample. Journal o!
rather t h a n directly on the m e a n and s t a n d a r d deviation of the Abnormal Child Psychology, 15. 309-327.
raw scores, t r a n s f o r m i n g the original d i s t r i b u t i o n to fit a nor- Hogan, A. E., Quay, H.C., Vaughn, S., & Shapiro, S. K. (1989). Re-
vised Behavior Problem Checklist: Stability, prevalence, and inci-
mal curve, with a m e a n of 50 a n d a s t a n d a r d deviation of 10.
dence of behavior problems in kindergarten and first-grade children.
From research findings ( L a F r e n i e r e & D u m a s , 1995 ) and clin-
Psychological Assessment, 1, 103-111.
ical experience, a n d in accord with m a n y i n s t r u m e n t s o f this lzard, C. E. ( 1977 ). ttuman emotions. New York: Plenum Press.
genre, we have found t h a t the lowest 10% of the children seem Izard, C. E. ( 1991 ). The ps)'cholo,g), q/emotions. New York: Plenum
to set themselves a p a r t from the rest o f the children in signifi- Press.
cant ways. Because a percentile r a n k o f 10% corresponds to a T Kohlberg, L., LaCrosse, J., & Ricks, D. (1972). The predictability of
score o f 37 (90% equals 6 3 T ) , it is r e c o m m e n d e d that 37T and adult mental health from childhood behavior. In B. B. Wolman (Ed.).
63T be used as the clinical cutting points in the process of form- Manual q/childpsychopathoh~g): New York: McGraw Hill.
ing interpretative hypotheses. At no t i m e should the SCBE be Kohn, M., & Rosman, B. L. (1972). A social competence scale and
symptom checklist for the preschool child: Factor dimensions, their
used as the sole basis for any clinical decisions regarding inter-
cross-instrument generality, and longitudinal persistence. D~qeh~p-
vention with the child. Rather, interpretations based on the pro- mental Psychok~gy 6, 430-444.
file should be viewed as hypotheses that require further corrob- LaFreniere, E J. (1996). Cooperation among peers as a conditional
oration from other data sources, as well as the exercise o f sound strategy: The influence of family ecology and kin relations, lmerna-
professional judgement. For more extensive i n f o r m a t i o n con- tional Journal q/Behavioral Development, 19( 1 ), 39-52.
cerning clinical applications, see LaFreniere and D u m a s LaFreniere, P. J., & Dumas, J. E. (1992). A transactional analysis of
(1995). early childhood anxiety and social withdrawal. Deveh~pment and Psy-
chopathology, 4. 385-402.
LaFreniere, E J., & Dumas, J. E. ( 1995 ). Social competence and behav-
References ior evaluation (preschool ed.). Los Angeles: Western Psychological
Services.
Achenbach, T. M., & Edelbrock, C. S. ( 1981 ). Behavioral problems and LaFreniere, P. J., Dumas, J., Capuano, E, & Dubeau, D. ( 1992 ). The
competencies reported by parents of normal and disturbed children development and validation of the preschool socio-affective profile.
aged four through sixteen. Monographs qf lhe Sociel y./br Research in Psychoh~gical Assessment: Journal ~71C~msulting and Clinical P.U'-
cho/ogl; 4, 442-450.
Child Do,elopment. 46 ( 1, Serial No. 188 ).
LaFreniere, E J., & Sroufe, k. A. ( 1985 ). Profiles of peer competence:
Asher, S. R., & Dodge (1986). Identifying children who are rejected by
Interrelations among measures, influence of social ecology, and rela-
their peers. Developmental Psycholog); 22, 444-449.
tion to attachment history. Deveh)pmenla/l{~vchoh)gy, 21.56-69.
Behar, L. (1977). The Preschool Behavior Questionnaire..hmrnal o!
Peterson, D.R. (1961). Behavior problems of middle childhood.
Abnormal Child Psychology, 5, 265-275. Journal q/E'onsulting Psychology, 25. 205-209.
Behar, L., & Stringfield, S. (1974). A behavior rating scale for the pre-
Plutchik, R. (1980). Emotion: ,4 p,wchoevohttionary synlhesi,s, New
school child. Developmental Psychology, 10. 601-610. York: Harper & Row.
Bowlby, J. (1980). Atlachment and loss: l bl. 3. Loss. New ~brk: Basic Plutchik, R. (19913. Emotions and evolution. In K.T. Strongman
Books. (Ed.), International review qf studief on emotion: l~bl. 1. New York:
Coie, J. D., Dodge, K. A., & Kupersmidt. J. B. (1990). Peer group be- Wile>~
havior and social status. In S. R. Asher & J. D. Coie (Eds.), Peer re- Quay, H. C. ( 1983 ). A dimensional approach to behavior disorder: The
jection in childhood. Origins, consequowes, and intervention ( pp. 17- Revised Behavior Problem Checklist. School Pfychoh~gy Reviews: 12,
593. New York: Cambridge University Press. 244-249.
Dumas, J. E. ( 1994 ). Conduct disorder. In R. J. Corsini (Ed.), Encyclo- Richman, N., Stevenson, J. S., & Graham, P. J. (1982). Preschool to
pedia q/psychoh)gy ( 2nd. ed.; Vol. 2, pp. 289-293 ). New York: Wiley. school: :4 behavioral study London: Academic Press.
Dumas, J. E.. & LaFreniere, P. J. ( 19933. Mother-child relationships Rutter, M., & Garmezy, N. ( 1983 ). Developmental psychopathology. In
as sources of support or stress: A comparison of competent, average, P. H. Mussen (Ed.), Handbook o[child psychology: l,bl 4. Socializa-
aggressive, and anxious dyads: ChiM Deveh)pment. 64. 1732-1754. lion, personality attd social development ( 4th ed.: pp. 776-91 I ). New
Dumas, J. E., LaFreniere, P. J., & Serketich, W. J. (1995). Balance of York: Wiley,
power: A transactional analysis of control in mother-child dyads in- Sroufe, L. A. ( 1996 ). Emotional development: The organization qlemo-
volving socially competent, aggressive, and anxious children. ,hmrnal tional l~li"in the early years. New York: Cambridge University Press.
o/,tbnormal Psychoh)gy 104, 104-113. Sroufe, L. A., Schork, E., Motti, E, Lawroski, N., & LaFreniere. E J.
Ekman, P. (1984). Expression and the nature of emotion, In K. R. ( 19843. The role of affect in social competence. In C. E. lzard, J.
Scherer & P. Ekman (Eds.), ,,tpproaches to emotion (pp. 319-344 ), Kagan, & R. Zajonc (Eds.), Emotions, cognition and behavior (pp.
Hillsdale, N J: Erlbaum. 289-319 ). Cambridge, England: Cambridge University Press.
Fischer, K. W., Shaver, P. R., & Carnochan, P. ( 19903. How emotions United States Bureau of the Census ( 1991 ). Statistical Abstract qlthe
develop and how they organise development. Cognition and Emotion, United States: 1991 ( 11 lth Edition). Washington, DC: U.S. Govern-
4, 81-127. ment Printing Office.
Hinshaw, S. P., Morrison, D. C., Carte, E. T., & Cornsweet, C. (1987). Waters, E.. & Sroufe, L. A. (1983). A developmental perspective on
Factorial dimensions of the Revised Behavior Problem Checklist: competence. Developrnoltal Revieu; 3, 79-97.
 SCBE-30 377

Appendix A

M e a n s a n d S t a n d a r d D e v i a t i o n s o f S C B E - 3 0 Scales
f o r A g e a n d G e n d e r G r o u p s : U.S. S a m p l e

Girls Boys

Scale and age group M SD M SD

Social Competence
3 year olds 35.51 7.59 32.30 9.05
4 year ol~ls 39.35 8.95 36.93 9.06
5 year olds 41.11 9.07 38.57 9.23
6 year olds 43.65 6.66 40.68 9.10
Anxiety-Withdrawal
3 year olds 23.26 7.88 24.10 6.72
4 year olds 22.89 7.86 22.52 7.50
5 year olds 20.95 7.44 21.59 8.10
6 year olds 20.06 7.07 20.48 7.67
Anger-Aggression
3 year olds 19.40 8.78 24.88 11.69
4 year olds 19.99 9.45 22.53 10.07
5 year olds 16.79 7.62 19.40 8.77
6 year olds 15.49 6.23 18.30 8.76

Note. SCBE-30 = Social Competence and Behavior Evaluation Scale: Short Form.

Appendix B

Means and Standard Deviations of SCBE-30 Scales

for Age a n d G e n d e r G r o u p s : Q u e b e c S a m p l e

Girls Boys

Age group M SD M SD

Social Competence
3 year olds 37.27 7.46 35.58 8.06
4 year olds 40.76 7.19 36.48 8.11
5 year olds 42.51 6.92 38.06 7.05
Anxiety-Withdrawal
3 year olds 20.69 6.54 20.89 6.26
4 year olds 19.79 5.72 20.19 6.90
5 year olds 20.43 6.49 21.10 6.94
Anger-Aggression
3 year olds 22.73 7.38 23.43 8.32
4 year olds 21.11 7.21 24.89 9.23
5 year olds 21.69 8.18 25.33 8.95

Note. SCBE-30 - Social Competence and Behavior Evaluation Scale: Short Form.

Received January 26, 1996

Revision received April 22, 1996
Accepted April 30, 1996 •