The nature of 7-year-old children's (n = 140) positive peer relations
was examined using a multi-method, multi-agent approach. A confirmator
y factor analysis suggested three factors clustered by reporting agent
and method: teachers' perceptions of the child's positive attributes,
observation of the child-peer interaction in three play settings, and
observation of the child-peer interaction in a teaching context. Simi
lar pathways were found for boys and girls. Once it was established th
at stable constructs could be developed to measure positive peer relat
ions, earlier parent-child characteristics were examined to assess the
ir utility in predicting children's positive :I friendships. Parental
scaffolding and warmth were measured at 18 months and 5 years of age.
Structural equation modeling suggests that positive parent interaction
s were positively related to later peer relations, but this relationsh
ip was significant only for boys. Possible explanations for this sex d
ifference are discussed.