Abstract: A growing body of research suggests the appropriateness of social analogues to such psychological constructions as the "Imatching law" and the "flaw of effect." Using such analogues and their appropriate social applications we are able to show that apparent situations of inequity or exchange imbalance are actually often the result of balancing mechanisms which operate when several reinforcement domains are simultaneously operative in social situations. Thus imbalances may only appearfor local reinforcement domains while they are balanced over the total social situation, rendering the total system equitable. Such considerations may explain the apparently pervasive nature of some inequitable social structures. A growing body of recent research evidence has emerged demonstrating that elementary behavioral principles may be applied to the explanation of social phenomena with a high degree of empirical success (e.g., Bavelas et al.; Emerson, a; Gray and von Broembsen; Gray et al., a; Hamblin, a, b; Molm and Wiggins). We shall examine some of these basic behavioral principles from experimental psychology-most notably the "matching law" (Herrnstein, a; Killeen; Rachlin) and the "law of effect" (Herrnstein, b)-and indicate the theoretical implications of extending such principles to the explanation of social phenomena. Our primary aim is to develop social analogues to individual behavioral processes which allow us to examine various aspects of behavior in multiperson settings. Most of the current literature in sociology and social psychology (see Homans, b) assumes the implicit or explicit working of basic processes involving reinforcement, punishment, discrimination, chaining, etc. However, a stumbling block appears when direct generaliza1 982 The TTniupvriitvu of mNrh Carolina Press. nnf7-77l2/ ,/0n0n 6-R2$0n.7n 156 This content downloaded from 207.46.13.166 on Thu, 07 Jul 2016 06:14:30 UTC All use subject to http://about.jstor.org/terms Explaining Exchange Imbalance / 157 tion to group situations is attempted, since various questions arise concerning the compounding of such variables and processes. THE CONCEPTUAL UNIT OF ANALYSIS Early attempts to generalize behavioral principles to social events took the individual as the conceptual unit of analysis; that is, behavioral processes were seen to effect change in the individual. Later Homans (a) and Emerson (a) specified the exchange relationship as the focus, and the behavioral processes were conceived to effect the exchange relationship. We propose to carry this one step further. In the pages to follow we construct a theory of social behavior based on the assumption that the group is subject to the principles of behavioral psychology. Consider a group as a source from which several behaviors can be emitted. In the simplest situation each member of the group can behave in some specified manner and these behaviors constitute the range of options from which the group can choose. Once the group has chosen a course of action, then the environment external to the group may provide stimuli of some general reinforcing nature. In this manner the group's choices, even though they are not generally made solely by individuals, but instead represent a group product, represent behaviors which may be subject to reinforcement in the usual manner. Behavioral psychology presents many theories about the incidence of behaviors and their strengthening or weakening through the presentation of subsequent stimuli (see Thorndike). Since it is the incidence of behaviors with which we are concerned such theories could logically be applied to the behaviors of groups, though, of course, they need not be empirically correct and must therefore be tested. Complications do arise with this approach, however, since behavioral principles must be operating on two (or more) levels simultaneously. The individual members of a group must also be subject to the principles of behavioral alteration while the group level processes occur. Thus the group, as a unit, must serve in an exchange relationship with its individual members to ensure that they behave appropriately in order to elicit group level reinforcement. Thus we see that individual action is necessarily affected by group (or internal) reinforcement and that this group reinforcement serves itself as a type of operant behavior which may be reinforced by the external environment. The group level processes, moreover, need not represent some central tendency of the individual level processes since individuals do not necessarily repeat reinforced behavior on the part of others, but, instead, provide the reinforcement for behaviors which they, themselves, might alternatively have engaged in. If the principles of behavioral psychology can be applied at the group level as we propose, then balancing features of such processes should also apply, such as the "matching law" and the "law of effect." This content downloaded from 207.46.13.166 on Thu, 07 Jul 2016 06:14:30 UTC All use subject to http://about.jstor.org/terms 158 / Social Forces Volume 61:1, September 1982 Thus we begin our discussion with consideration of these principles and their implications for single organism exchange relations. We shall then proceed to generalize these principles to the group level and consider domains of reinforcement which should apply at that level. Specific application will be made in the development of a theory of social power. Exchange Implications of Individual Processes VARIETIES OF THE MATCHING LAW Psychologists have long been interested in the relation between a response measure such as number of responses, proportion of responses, or time spent in response, and the frequency, probability, or rate of reinforcing stimuli (see Estes, a; Skinner; Thorndike). Interest in the reinforcing stimuli has divided over the question of whether it is the received reinforcing stimuli or the scheduled reinforcing stimuli that is most important in determining response with adherents of operant conditioning favoring the received reinforcing stimuli acquired during application of interval and ratio schedules of numerous varieties. Estes (a, b) developed a simple model for non-contingent learning (reinforcement is not dependent on behavior) which, at asymptote, predicts that