My research focuses on the problem solving and decision-making performance of small groups. Broadly stated, this work examines the interpersonal and social interaction processes by which small groups of people transform a diverse array of facts, opinions, and preferences into consensual, group-level judgments and decisions. This research seeks to better understand how groups handle the information that is available to them, and how their information-handling capabilities affect the decisions they make.

I take an information-processing approach to the study of group problem solving and decision making, although at the group level, information processing is conceptualized in distinctly social psychological terms. This approach is described more fully in Larson and Christensen (1993), which examines in detail the relevance of social interaction for, among other things, the acquisition, storage, transmission, manipulation, and use of information for the purpose of creating group-level judgments and decisions. For related theoretical treatments, see Christensen and Larson (1993), Fiske & Goodwin, 1994, and Hinsz, Tindale, and Volrath (1997).

Building on this conceptual foundation, a major thrust of the empirical research that my students and I conduct has been to understand how groups are affected by an unequal distribution of information across members. Problem solving and decision-making groups often consist of individuals who possess overlapping, but not identical, subsets of information concerning the problem or decision at hand. Thus, members often hold unique (unshared) information that others in their group do not possess. This stands in contrast to the shared information they hold in common with many group members. The distinction between shared and unshared information raises a number of interesting research questions.

One set of questions concerns the degree to which groups discuss shared versus unshared information. There is substantial evidence from my laboratory and elsewhere (for a review, see Wittenbaum & Park, 2001) that groups often discuss more of their shared than their unshared information, and that they discuss their shared information more thoroughly. One aim of our research has been to document the time course of this phenomenon. Toward this end we have developed and tested a dynamic information sampling model of group discussion. A detailed description of this model can be found in Larson, (1997). Not only does this model account for the tendency of groups to discuss more shared than unshared information overall, it predicts that groups will tend to discuss their shared information earlier than their unshared information. This model has been tested in our laboratory with ad hoc groups of undergraduate psychology students performing hypothetical decision-making tasks (Larson, Foster-Fishman, & Keys, 1994; Larson, Foster-Fishman, & Franz, 1998, Winquist & Larson, 1998), and in the field with teams of practicing physicians diagnosing patient medical cases (Christensen, Larson, Abbott, Ardolino, Franz, & Pfeiffer, 2000, Larson, Christensen, Abbott, & Franz, 1996; Larson, Christensen, Franz, & Abbott, 1998). The results from this research provide substantial support for the dynamic information sampling model. In a recent off-shoot of this work, we have been exploring how the model might be modified to account for another interesting finding: the tendency of groups to repeat more of their already-mentioned shared than unshared information (Larson & Harmon, 2007).

Another set of questions concerns the role that group leaders play during problem solving and decision-making discussions. Specifically, we have sought to determine whether leaders are able to counteract the tendency of groups to discuss their shared information more thoroughly than their unshared information. Our research has shown, for example, that an appointed leader is more likely than other members of the group to play an information management role during group discussion (e.g., Larson, Christensen, Abbott, & Franz, 1996; Larson, Christensen, Franz, & Abbott, 1998). Among other things, this means that leaders are more likely than other members to draw the group's attention back to information brought up earlier in discussion. This leader/member difference is most apparent with respect to the shared information that gets mentioned, but also occurs for unshared information, especially toward the end of discussion. We have also shown that leadership style can significantly affect the balance of shared versus unshared information that gets discussed (Larson, Foster-Fishman, & Franz, 1998). In comparison to directive leaders, participative leaders seem better able to get group members to discuss their unshared information. Interestingly, however, groups led by participative leaders do not necessarily make better decisions than groups led by directive leaders.

A third set of research questions has to do with whether the tendency to discuss more shared than unshared information is detrimental to the overall quality of the decisions that groups eventually make. The answer is complex, and depends in part on whether the information pointing to the objectively best choice alternative is shared or unshared prior to discussion. If it is unshared, then a sub-optimal choice alternative is apt to be selected by the group. We have found evidence of this among novice decision makers (e.g., Winquist & Larson, 1998) as well as among highly trained professionals working on decision-making tasks similar to those they perform every day (Christensen, Larson, Abbott, Ardolino, Franz, & Pfeiffer, 2000; Larson, Christensen, Franz, & Abbott, 1998). Further, compared to groups led by a participative leader, those led by a directive leader also tend to select a sub-optimal choice alternative when the information pointing to the objectively best choice alternative is initially unshared. But this occurs primarily when the directive leader personally prefers the sub-optimal choice alternative. On the other hand, when the directive leader personally prefers the optimal choice alternative, his/her group actually is more likely than a group led by a participative leader to select the better choice alternative (Larson, Foster-Fishman, & Franz, 1998).

A fourth set of questions concerns whether shared information confers greater influence on group members who hold that information. One implication of the findings reported in the preceding paragraph is that shared information tends to have a stronger influence on group decision making than does unshared information. If so, and if some members hold more shared information than others, it seems possible that those members themselves may acquire greater influence within the group (but see Larson, Sargis, Elstein, & Schwartz, 2002). For example, in one recent study of dyadic preference factions within four-person groups, we demonstrated that factions whose members held much information in common with one another prior to discussion (i.e., shared information) had greater influence on the group's eventual decision than did factions whose members held little information in common with one another (Larson, Sargis, & Bauman, 2004). It is likely that this occurred because preference factions whose members held much information in common with one another were able to work together more effectively to construct strong, persuasive arguments in support of their decision preference (see also Sargis & Larson, 2002). These findings and others serve to illustrate how certain older, more traditional concerns within the field of social psychology (group structure, subgroup influence) may be profitably informed by a contemporary analysis of the informational dynamics within groups.

Three other recent projects are worth noting. One is a just-published book titled "In Search of Synergy in Small Group Performance" (Larson, 2010). This book critically evaluates more than a century of empirical research on the effectiveness of small, task-performing groups, and offers a fresh look at the costs and benefits of collaborative work arrangements. The central question taken up in the book is whether, and under what conditions, interaction among group members leads to better performance than would otherwise be achieved simply by combining the separate efforts of an equal number of people who work independently. The book considers this question with respect to a range of tasks (idea-generation, problem solving, judgment, and decision-making) and from several different process perspectives (learning and memory, motivation, and member diversity). As a framework for assessing the empirical literature, I introduce in the book the concept of synergy. Synergy refers to an objective gain in performance that is attributable to group interaction. Further, I distinguish between weak and strong synergy, which are performance gains of different magnitude. The book highlights the currently available empirical evidence for both weak and strong synergy, identifies the conditions that seem necessary to produce each, and suggests where the search for synergy might best be directed in the future.

A second project is an agent-based model of group problem solving (Larson, 2007). An agent-based model is a computer program that simulates simultaneously multiple agents, or actors, who behave in ways that impact one another either directly or indirectly. In this particular model I try to express in programming code some rather simple theoretical ideas about (a) individual problem-solving, (b) group composition, and (c) group interaction. When expressed in this way, these ideas can be "run" (by running the program) to see what implications they have for problem solving in groups. A particular focus of this work has been the effect of diversity in members' problem-solving styles on group problem-solving performance. To date, the most significant prediction derived from this project is that when members are similar to one another in problem-solving style, the group as a whole is apt to perform better that its average member, but not as well as its best member. On the other hand, when members have very different problem-solving styles, the group as a whole can perform better than even its best member. Or, in terms of the synergy concept I introduce in the book (see preceding paragraph), groups that are homogeneous with respect to their problem-solving style are predicted to exhibit weak synergy, whereas groups that are diverse with respect to their problem-solving style are predicted to exhibit strong synergy. For a non-technical overview of how computers can be made to simulate group behavior, and how such simulations can benefit theory development, see Larson (in press).

Finally, a project that I am just beginning (and about which little is yet written) has to do with motivation gains in group work settings. An enduring question in the field of group dynamics is whether people exert any more or less effort when they work on tasks as part of a group in comparison to when they work alone. Twenty years ago, the textbook answer to this question was that people routinely exert less effort when working in groups. That conclusion reflected the research literature at the time, which was dominated by studies of social loafing and free-riding in groups. Both are phenomena that involve withholding effort in collective work situations, and both suggest that when performing tasks as part of a group, people often do not work as hard as they otherwise would have worked had they engaged in exactly the same activity completely on their own. During the past two decades, however, a more nuanced picture has begun to emerge, one that allows for the possibility that people may sometimes exert more effort when working in groups than when working alone. To date, the evidence for this comes almost exclusively from the least capable members of groups that are asked to perform unitary conjunctive tasks. A unitary conjunctive task is one in which all members must contribute identically to the group's product, with the implication that the group's collective performance can be no better than that of its least able member. An example is a mountain climbing team whose members are roped together while making their assent. Such teams can climb at a pace that is no faster than their slowest members are able to maintain. Research demonstrates that on tasks of this sort, the least capable members typically exert more effort than they would have exerted had they performed exactly the same activity working completely alone (for recent reviews, see Larson, 2010; Weber & Hertel, 2007). Although this effect has so far been observed only among those who are the weakest or least capable members of their group, a careful theoretical analysis suggests that with a slightly change in the underlying nature of the task being performed (i.e., so that it conforms to what I call a divisible conjunctive task) the same effect should appear as well among those who are the strongest and most capable members of their groups. I am currently explore this idea in a pilot study using a specially designed physical persistence task that I hope will become a testbed for a series of controlled experiments to be conducted over the course of next several semesters.