|Abstract or Summary
- Social competence has emerged in research on social development
as a construct unifying behavioral and motivational dimensions of social
development in the preschool child. It draws from theories of social
learning, cognitive development, and competence motivation. The
theoretical relevance and empirical utility of the construct indicate
the need for further research on methods of measuring social competence.
Recent investigators have noted the cultural relativity of
social competence, as a construct which includes values about behavior.
Although parents have been infrequently used as a population of experts
for content validation, they could provide important information about
the cultural values underlying a definition of competence.
The present study investigates several methodological questions
as they relate to Baumrind's Preschool Behavior Q Sort (PBQS), a measure
of social competence. The questions include reliability and validity
after modifications in administration; the construct validity of the
PBQS factor Independence, when compared with an alternative measure of
independence; and the content validity of the PBQS seven-cluster, two-factor
model of preschool competence based on parental responses to
Child data came from naturalistic observations of 36 preschool
The sample was restricted to children between 48 and 60 months old, of
normal and above verbal intelligence, and from middle class families.
The PBQS was used to rate subjects after a 2.5 hour observation supplemented
with anecdotal records collected by staff at the preschool.
Parent subjects had children in the preschool which provided the
child sample. The parent sample was homogeneous with respect to sociocultural
background. Parents rated PBQS items presented as seven point
Likert-type items. Questionnaires were completed by 97 parents.
Interrater reliability was computed for each subject across 72 items
and 7 clusters, for each cluster across all subjects, and for each item
across all subjects. Reliability was high for most subjects. Reliability
was significant for all clusters and 63 items.
The matrix of interitem correlations for child data was analyzed on
the basis of frequency of significant correlations and the mean correlation
of each item with all others. Although some within-cluster correlations
were in the expected pattern, between-cluster item correlations indicated
a lack of independence between clusters.
A stepwise discriminant analysis was used to test the relationship
of the PBQS Independence factor to another measure of independence. In
a model which discriminated between children having three levels of independence
on Beller's Scale of Independence, less than one-half the
items were PBQS Independence factor items.
The matrix of interitem correlations for parent data was analyzed
in the same way as the matrix for child data. The multi-cluster pattern
of intercorrelations did not appear. Parents seemed to respond in terms
of a univariate definition of social competence, with a strong consensus
on items describing cooperation and compliance. There was a pronounced
absence of consensus on most items describing independence.
The modified administration procedure for the PBOS appeared to produce
reliable ratings with questionable validity. Therefore, the results
and conclusions of the study were applied to modify the cluster
scoring procedure of the PBQS. The resulting three-cluster model appears
to be more simplistic than Baumrind's seven clusters, but it is
more valid in terms of the child ratings for the present study. The
deviation of the three-cluster model from a parental definition of competence
is easily delineated, since one of the three clusters strongly
represents the parental definition for the sample of parents used in
In view of the empirical method by which clusters were derived,
limited clinical or theoretical significance can be attached to the
clusters. The clusters represent functionally related behavior indices,
but their structural relationship has not been demonstrated.