Abstract
Existing research suggests that the nature of the input data can significantly affect the character of the outcomes. In our study, we examined differences in the social structure of children's peer groups based on sociometric and observational data collection methods and for children of different ages.
Four classes of preschool, four classes of prepubertal, and two classes of pubertal children were included. Sociometric data were obtained through interviews with children about their friendship with classmates.
Observational data were obtained using the behavioral interaction scanning method simultaneously with interval scanning of the spatial configuration of the children in the classroom during free play. Based on the collected data, three social networks (friendship nomination, affiliative interaction and proximity) were generated for each class.
Strength (the number and strength of social ties of individual children) and Transitivity (the degree to which individual children distribute their social ties evenly across classmates) were subtracted from each network. Gender assortment and group hierarchy were also calculated for each network.
The analysis showed that most of the known phenomena, such as more friends for girls and smaller close-knit groups for girls, are based on children's friendship reports but not on observed behavior. Results also showed higher gender assortment (across all ages) and hierarchy (in preschoolers and prepubescent) in friendship nomination networks than in affiliative interaction and proximity networks.
In summary children's friendships perceptions are not necessarily reflected in their actual behavior. Accordingly, different data collection methods may generate different structural group arrangements and yield different conclusions.