This investigation focused on the friendship concepts of pre-school-aged children. Eighty-four 3-to 5-year-old children responded to questions about social provisions, or benefits of relationships, available from reciprocal friends (both children in a pair nominate each other as friends), unilateral friends (one child nominates the other), non-friends (neither child of a pair nominates the other), and imaginary friends ( for those who had them). Results indicated that reciprocal and imaginary friends were the best sources of social provisions, followed by unilateral friends and non-friends. Major implications are that children draw distinctions between reciprocal and unilateral friends based on the levels of social provisions available in these relationships, and that relationship schemas underlying relationships with imaginary companions may be akin to those for reciprocal friends. Children's friendship schemata may also differ by gender, as girls discriminated more than boys between reciprocal and unilateral friends for the social provision of companionship.

The influence of three components of temperament (activity level, impulsivity, and soothability) on children's friendships was investigated. Children (40 girls, 35 boys) aged 43 to 69 months responded to a sociometric interview and teachers provided temperament ratings. The probability of children choosing particular classmates as friends was evaluated based on the genders and temperaments of the dyad. A logistic choice model revealed that the choice of friends is highly influenced by gender, high impulsivity, and high soothability. Furthermore, the gender of the chooser and the activity level of the friend interacted such that girls chose low activity level friends and boys chose high activity level friends. In addition, the likelihood of a child being chosen as a friend based on gender and temperament was significantly correlated with popularity for girls, but not for boys.

The influence of three components of temperament (activity level, impulsivity, and soothability) on children's friendships was investigated. Children (40 girls, 35 boys) aged 43 to 69 months responded to a sociometric interview and teachers provided temperament ratings. The probability of children choosing particular classmates as friends was evaluated based on the genders and temperaments of the dyad. A logistic choice model revealed that the choice of friends is highly influenced by gender, high impulsivity, and high soothability. Furthermore, the gender of the chooser and the activity level of the friend interacted such that girls chose low activity level friends and boys chose high activity level friends. In addition, the likelihood of a child being chosen as a friend based on gender and temperament was significantly correlated with popularity for girls, but not for boys.