Directed network analysis of 2-year-old and 4-year-old children

This dataset includes triaxial acceleration data collected from 60 children (2- and 4-year-olds) across three childcare facilities during a 15-minute free play session. Dyadic (two-person) and triadic (three-person) peer relationships were analyzed using a novel-directed network analysis method, offering insights into age and sex differences in peer interactions. The dataset includes normalized connection counts, demographic details, and detailed analyses of interaction directionality. This research validates the application of network analysis in early childhood studies, reducing observational biases and labor-intensive manual coding, and provides a framework for exploring complex social dynamics in naturalistic play settings.

Methods
Participants:
The study involved 60 children, with equal representation of 2- and 4-year-olds, across three childcare facilities. Informed consent was obtained from the participants' legal guardians following ethical guidelines.

Data Collection:
Participants wore wristwatch-style triaxial accelerometers (Silmee W22, TDK) during a 15-minute free play session. Acceleration data were recorded at 20 Hz to measure individual movement intensity and calculate dyadic and triadic peer relationships.

Data Processing:

Normalization: Acceleration data were aggregated into 1-second intervals and normalized using Sturges’ formula to account for individual variability.

Network Analysis: Connections were quantified using directed graph analysis, identifying dyads and triads based on movement entropy thresholds. Entropy values ranged from 0 to 1, representing interaction strength.

Statistical Analysis:
Chi-square tests and ANOVAs were performed to analyze age, sex, and directional differences in peer relationships. Data standardization addressed variations in potential pairings across groups.

Tools and Software:
Analysis was conducted using IBM SPSS (ver. 29.0) and a custom network analysis system developed by Hitachi Ltd.