The study presents the results of testing an original obersvation instrument designed to identify a power elite in multiple sessions of two eight-member adult discussion groups. Two questions guided the development of the study: (1) Can a power elite be identified in participation training groups? (2) Can an effective observation instrument with acceptable technical characteristics be developed to identify behavior manifestations of power in participation training groups? Related literature is reviewed, testing methodology and statistical procedures discussed. The study concludes that both questions are answered affirmatively and that in support of statements made that power relations are an integral part of group processes in which decisions are made, power identification is a starting place for assessing where a group is in relation to patterns of influence in decision-making and problem-solving. A bibliography, tables, and social interaction survey are included. (MW)