46
randomly selected two five-minute segments of videotape for each dyad to re-code. Reliability estimates had already been obtained for 30% of the sample from which the clinic-referred families were drawn.