ABSTRACT
An observation system yielding information on children's social behaviours in free play settings together with procedures for training observers to achieve reliable data is described. Assessment of social involvement in play activities using the Parten Scale is combined with frequency and topographic information on social interactions to describe individual child behaviour while simultaneous observation of three children gives comparison data against which to evaluate behaviours of a target child. Results from observations of three children in each of three kindergartens over twelve sessions showed high interobserver reliability with validity indicated by data discriminating teacher identified low socially involved children from randomly selected comparison children. Sensitivity to the effects of intervention was also shown and examples given of the use of the procedures in clinical assessment. Data from an evaluation of the effects of priority coding versus recording of all codes in each interval showed priority coding to give conservative data on social involvement in play, and this must be set against additional observer training necessitated by the more complex procedures when using the system.
