On the basis of behavioral criteria the study conducted a cross-sectio
n comparison of coercive patterns by three and four year olds identifi
ed as agressive or non agressive. Behavior was recorded in two setting
s: classroom and playground. Trained observers used the Behavioral Obs
ervation of Social Interaction system achieving reliability superior t
o 79%. The Kappa concordance analysis revealed excellent levels on the
basis of the criteria proposed by Bakeman & Gottman (1986). In genera
l, the behavioral criteria for subject selection allowed for the diffe
rential identification of coercive interactive patterns between childr
en classified as agressive and those from a comparison group. Younger
preschool children showed more coercive behavior than older preschoole
rs and those from the comparison group.