Monday, October 4, 2010

Improving group performance: Recent research results

Conversational turn-taking—as opposed to having a few people dominate the conversation—wins the prize for the most effective way to increase a group's intelligence. Research reported September 30, 2010 in Science defines collective intelligence as "the general ability of the group to perform a wide variety of tasks" and shows that three factors contribute to it: conversational turn-taking, social sensitivity, and proportion of females in the group. Since women scored better than men on social sensitivity, and social sensitivity could reasonably lead group members to allow others to speak, I am prepared to conclude that conversational turn-taking is the key. Read more about the study here. If you're interested in research on improving group performance, please get in touch.


  1. I see these three factors as (a) related, (b) not reducible to one of them, and (c) not the whole story. It seems obvious to me that all three could theoretically be present and the group not generate collective intelligence -- especially if there is not a commensurate ability of those involved to deeply HEAR each other. I'd put "people feeling fully heard" as the most potent factor.

    But there's more to it than that. The definition is very strange. You can build a machine that can "perform a wide variety of tasks" but that wouldn't necessarily make it intelligent. Intelligence has to do with the ability to create, apply, and modify our mental models and narratives so that they align with (and align us with) the real, changing world we live in. The test of real intelligence is the ability to engage successfully with novel circumstances. I think the factors listed by the SCIENCE article play significant roles in that, but much more is needed. A major factor unmentioned is the need for high quality information (which is often enhanced, but not ensured, by people taking turns and listening to each other -- but the diversity of the group can be just as important... etc.)

    We also need to explore collective intelligence not just in groups, but in organizations, networks, communities, societies, and humanity as a whole. What are the factors that determine our functional intelligence THERE? That inquiry moves us into really critical territory

    What excites me here is not the findings of the research (which barely scratch the surface of the subject of CI), but the fact that such research is now being done formally and published in reputable major scientific journals like SCIENCE. I say: Let's see LOTS MORE of such research. The field is virgin territory for it, and the world desperately needs much much greater understanding of how to generate collective intelligence -- and collective wisdom. In fact our collective lives may quite literally depend on it.

  2. Your points are well taken. I will admit to being somewhat reductionist in summarizing the results of this study.
    The full research article and supplemental materials are available at

    Here is some supplementary information.

    The researchers used McGrath's "circumplex of group tasks" (which includes four basic types of tasks--idea generation, choosing among alternatives, negotiating a solution to a conflict, and executing a task)* to ensure that they included some minimum variety of tasks. This is certainly preferable to using only one task, or multiple tasks of a similar nature. Nonetheless your point is well taken, and there is certainly more variety and complexity of tasks to explore.

    I would agree that "people feeling fully heard" is important. It seems to me to be consistent with the "conversational turn-taking" findings of this research. Although "conversational turn-taking" is more superficial, it is more objectively measurable. A subjective measure of being heard or understood would be a useful addition.

    A major limitation of the study was its use of randomly selected participants, whereas in practice, participants would (hopefully) be selected based on their diversity of pertinent knowledge, expertise and points of view. Further, as randomly selected strangers, they had no interpersonal or group history, so trust or personal stakes in outcomes were not factors. As is often the case with experimental research, generalizability is an issue.

    Also, I would agree that more research focused on collective intelligence would be valuable. Further, it would be valuable if we took advantage of the existing research in this area, which might be categorized in other terms--such as "group performance" or "group effectiveness"--rather than "collective intelligence."

    *McGrath, J. E. (1984). Groups: Interaction and Performance. Prentice Hall.

  3. Honeybee Democracy

    Here are some interesting findings about how honeybees make decisions, and their potential application to human groups. The following is excerpted from a recent book review.(1)

    “Human groups … could learn from how bees make decisions … minimizing a leader's influence, allowing each group member to contribute their opinions in an independent and unbiased manner, and only reaching a single group decision once a democratic quorum has been reached.”

    While this advice seems sound, the authors note some important differences.

    “Human groups are frequently not united by common interest in the way that honeybee swarms are united by shared kinship. The former often comprise conflicting factions each fighting for their own self-interest. And when human groups do act as cohesive units, they are often too cohesive, with their members rarely acting as independent decision-makers like honeybee scouts. Conformity prevents dissenting views and conflicting evidence from being considered … Whereas honeybee swarms are cooperative by virtue of shared kinship, groups of people are cooperative partly through conformity. Eliminating conformity may eliminate decision-making errors, but it may also reduce the cohesiveness that maintains human groups in the first place.”

    (1) “Insect Swarm Intelligence,” by Lars Chittka and Alex Mesoudi, Science 28 January 2011: vol. 331 no. 6016 pp. 401-402, a book review of Honeybee Democracy by Thomas D. Seeley, Princeton University Press, Princeton, NJ, 2010.

  4. Bird Song Yields a New Understanding of Cooperation

    "Perhaps in human endeavors it is more important to have an image of what a group wants to achieve than each participant's own tasks."

  5. Large Group Interventions: An Empirical Field Study of Their Composition, Process, and Outcomes

    by Christopher G. Worley, Susan A. Mohrman, and Jennifer A. Nevitt

    Journal of Applied Behavioral Science 2011;47 404-431


    Large group interventions are an important method of organization change. The large group intervention literature is largely descriptive and normative and contains a number of case studies that describe the process and some immediate outcomes. There is a large void with respect to empirical investigation. This research tested fundamental hypotheses related to large group composition, process, and outcomes in a field study. Six large group interventions (decision accelerators) were used to develop clinical service area strategies and instigate strategic change in a health care system. The results support the assertion that stakeholder diversity in the group’s composition affects the number of stakeholder perspectives that were heard during the meeting and the breadth of issues addressed during decision making, but failed to support the assertion that composition affects the intensity of debate and disagreement. Stakeholder diversity had a weak relationship with novel and relevant large group outcomes, but debate intensity was strongly related to those outcomes. The implications of these results on large group intervention research and practice are discussed.


Any comments?