This paper reviews 31 empirical studies of small groups in which the major independent variable, group size, was related to several classes of dependent variables: group performance, distribution of participation, the nature of interaction, group organization, member performance, conformity and consensus, and member satisfaction. Many of these variables were found to be significantly affected by group size, but methodological shortcomings characterizing this group of studies preclude the assertion of broad generalizations. Several dependable and nondependable intervening variables are suggested which may help to account for many of the observed effects. Conclusions are: group size is an important variable which should be taken into account in any theory of group behavior, and future research on group size should proceed more systematically than in the past.

No abstract

VIEWS of the research on group problem solving (Kelley & Thibaut, 1954; Lorge, Fox, Davitz, & Brenner, 1958) indicate that sweeping conclusions about the superiority of groups over individuals are not warranted. However, studies in this area show that it may be fruitful to examine the conditions under which groups should be superior to individuals. This paper examines conceptually the requirements of the problem solving task and selected group variables that affect the quality of a group's solution, and suggests three models for group problem solving. Following the presentation of an experiment that tests some of the assumptions of the models and sheds further light on processes affecting the quality of group solutions, the paper terminates with a discussion of other models of group problem solving and of extensions of the models to preference phenomena.Each model specifies theoretical probabilities for the outcome of group problem solving, and the antecedent conditions that should produce the expected outcomes. A significant restriction on the application of the models is that the group's product must be expressed as the distribution of individual answers. Distributions of individual responses as group products are found in legislative votes, committee votes, and in court decisions, although these are not always answers to problems. A less imposing restriction is that the individual answers must be expressed as one of three or more discrete possibilities, such as a correct answer versus two or more incorrect answers; this restriction will be relaxed at a later point in the paper.The independence model states the theoreti-1 This article reports portions of a research project financed by a grant from the Horace H. Rackham School of Graduate Studies of the University of Michigan.2 The authors wish to express their gratitude for the help provided by Rosemary Conzemius in the early phases of this study.

Changing emphases in programs of social welfare, including public health, make it increasingly necessary to plan and conduct programs responsive to the subjective needs and values of the consumer groups for whom they are intended. Obtaining information about such needs in turn requires the development of inexpensive but valid tools for collecting data.In recent years the telephone has seen increasing use as a means of collecting data quickly and inexpensively. Despite some limitations, a small number of published studies generally support the usefulness of the telephone as a means of obtaining personal information. Unfortunately, most of these studies have focused on a relatively small geographic area, where the cost of telephoning is minimal and where it is feasible to undertake intensive efforts to reduce nonresponse rates. The present report is one of the very few which attempted to obtain personal information via the telephone from a probability sample of the U. S. adult population. It also provides detailed data on the costs of this method, and on nonresponse rates, and estimates possible sample bias stemming from differences in personal characteristics of telephone owners and nonowners. The results of an effort to reduce bias are also described. Finally, since the sample interviewed by telephone had earlier given personal interviews, data are provided on the consistency of information obtained by the two methods. THE STUDY SETTING

&dquo;Conflict management&dquo; usually refers to purposeful control of the overt behavior of parties to a conflict, such that their interactions remain within certain qualitative and quantitative limits. The most frequently specified limits imply the prevention or minimization of violence, but this is not essential to the defintion. Control of the psychological states of the parties (e.g., prevention or minimization of hostility) is often implied, but this is usually seen as one means of controlling behavior, rather than as the main goal. By definition, then, successful conflict management depends on successful control of behavior.Efforts to control behavior are virtually universal, since most human individuals and groups make frequent or continued efforts to control both their own behavior and the behavior of others. A large proportion of these efforts occurs in or generates conflict situations, and is made by the parties to the conflict as well as by &dquo;interested&dquo; third parties. It is therefore logically possible for anyone, regardless of his or her role in a particular conflict, to engage in efforts at conflict management.Unless the &dquo;manager&dquo; is merely lucky, successful behavior control would seem to depend on two things: knowledge of the at Monash University on April 11, 2015 abs.sagepub.com Downloaded from