Abstract
The present study attempted to evaluate the effect on children's behaviour of two key components of changes in group size—the addition of an extra group member and the role of that additional member. To this end, comparisons were made between adult-child dyads, adult-child-child triads, and adult-adult-child triads. Results suggest that the specific identity of the third person matters comparatively little: effects were similar regardless of whether the third person was an adult or a child. The implications of this for understanding second order effects and polyadic interaction are discussed.