This article reports the results of two studies examining some factors that increase the likelihood that employees will voice to their supervisors. The way employees perceive that their supervisors manage employee voice was identified as a major cause of the likelihood that employees will voice upward. The Supervisor as Voice Manager Scale is presented along with data demonstrating its reliability and validity. Theoretical and applied implications of the results are discussed.

This study develops a typology of decision paths through analysis of47 recorded decisions. The sample included a wide variety of decisions made by a variety of groups. Recent research has questioned the traditional assumption that all normal groups follow a set, unitary sequence of decision stages. Instead, this research suggests groups may follow multiple sequences in making decisions. Some writers have gone so far as to deny the existence of phases and propose alternative nonphasic formulations. This study sought to generate descriptive data that could hefp define the range of alternative decision paths and their properties, in the hope that this coufd helpsefect among the three perspectives on group decision development. The study developed new methods for mapping developmentalsequences that made itpossible to auoidoneofthemost troubling problems in prior research-the use of arbitrary time segments to definephases.The resulting typology had 11 different decision paths that fell into three main groups: unitary sequences, complex sequences, and solution-centeredsequences. Analysis of this typology supported the multiple sequence model, but also suggested a role for normative unitary models. The study also found that most groups have significant periods of disorganized activity. 0 ne of the most useful strategies for the study of group decision making focuses on the development of decisions over time. Over the past thirty-five years more than a hundred developmental studies have appeared, and scores more have invoked developmental assumptions. This interest reflects the widely accepted premise that we must study the process by which decisions are made if we are to understand and improve them (Fisher, 1975; McGrath 81 Altman, 1966).

This study tested a model of the factors that influence groups to follow various paths as they make decisions. Three panels of contingency variables, which indexed the nature of the group's task and group structure, were used to predict properties of the group's developmental path. The decision path properties included type of decision path, the complexity of the path, the amount of disorganized behavior the group engagedin, and the relative emphasis on various types of decision activity, such as orientation, problem analysis, and solution development. The contingency variables were quite effective in predicting decision paths and their properties. Group structure variables were stronger predictors than were task variables. Resultssuggested that significant revisions in the original contingency model were necessary. Most notably, the groups appeared to be much more rational in adapting their paths to the contingencies than we had originally supposed they would be.here is mounting evidence that group decisions do not develop in a single, universal set of stages. Rather, their T evolution may follow any of several alternative sequences. 549 550 HUMAN COMMUNICATION RESEARCH / Summer 1989 Multiple sequences have been found for a wide variety of decisions, from day-to-day operating decisions (Chandler, ). The previous study from this series (Poole & Roth, 1989) developed a typology of 11 distinct decision paths based on analysis of 47 operating decisions. In other arenas, Nutt has identified five paths for planning processes, and Mintzberg et al., seven paths for strategic decision-making processes. While there are several typologies of decision paths it is not yet clear what factors influence groups to take various paths. This study attempted to test a contingency model of decision development proposed by Poole (1983b). Three panels of contingency variables, which indexed the nature of the group's task and its internal structure, were used to predict properties of developmental sequences from Poole and Roth's (1989) typology.Poole's model allows for the influence of normative decision models and contingency factors on group decision activity (Poole, 1983b;Poole & Doelger, 1986). The model assumes that groups actively structure their decisions, as opposed to the often implicit assumption that group behavior is a reaction to external, determining causal factors. The following picture of group decision making emerges.Group members attempt to adapt their activities to the contingencies of the decision task, their current relationships with other group members, and the demands of the group's environment so as to make the most effective decision possible. Among the "tools" they use in this adaptive process are idealized, normative models of decision making such as those portrayed in traditional unitary models of group decision development. These models-part of the stock of knowledge every competent member of society has-organize the group's behavior in a general way. They serve as frames to help members make sense of their...

Using theoryfrom both small group and conflict management research, this study examined how conflict is managed within work groups performing decision-making tasks. It was proposed that larger and more cohesive groups would show less competing and more avoidance, compromise, and accommodation in response to conflict. We assessed whether conflict would be handled differently depending on whether the group is focused on problem analysis, solution evaluation, or group process management, and whetherconflict handling would change as afunction of time. Contingency analyses of 25 group decisions content-codedfor conflict-handling behavior, task achievementfunction, group size, cohesiveness, and temporalphase indicated mixed supportforour expectations. Conflict-handling behaviors reflecting high concern for others occurred mostfrequently, whereas behaviors representative of low concern for others occurred leastfrequently. Results are discussed in terms of hypothesis generation and suggestionsforfuture research thatcould continue linking these two literatures.